Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
II
Institut für Verfahrenstechnik der
Technischen Universität München
Prediction of Adsorption Equilibria of Gases
Ugur Akgün
Vollständiger Abdruck der von der Fakultät für Maschinenwesen der
Technischen Universität München zur Erlangung des akademischen Grades eines
Doktor-Ingenieurs
genehmigten Dissertation.
Vorsitzender: Univ.-Prof. Dr.-Ing. Weuster-Botz
Prüfer der Dissertation: 1. Univ.-Prof. Dr.-Ing. A. Mersmann, emeritiert
2. Univ.-Prof. (komm. L.) Dr.-Ing. J. Stichlmair, emeritiert
Die Dissertation wurde am 25.10.2006 bei der Technischen Universität München
eingereicht und durch die Fakultät für Maschinenwesen am 21.12.2006 angenommen.
III
Acknowledgments I would like to express my deep and sincere gratitude to my supervisor, Professor
Dr.-Ing. Alfons Mersmann, em. His wide knowledge and his logical way of thinking
were of great value for me. His understanding, encouraging and personal guidance
provided an ideal basis for the present thesis.
I am deeply grateful to Professor Stichlmair, who was member of my committee.
My sincere thanks are to Professor Weuster-Botz for acting as chairman of my
examination. I especially thank Professor Lercher and Dr. Staudt for the help
extended to me and the valuable discussion that I had with them during my research.
Special thanks are to Dr.-Ing. PhD Walter Roith for his guidance. Only people like
him, who have walked the same rocky way, can understand the efforts we made and
the happiness we received.
For giving me the impetus to restart the thesis after a break of several years I
would like to thank Dr.-Ing. Axel Eble.
I am grateful to Dr.-Ing. Lutz Vogel for the experiences we shared on the bus
route 53 in Munich, Dr.-Ing. Björn Braun for Schafkopfen, Dr.-Ing. Frank Stenger for
listening, Dr.-Ing. Lydia Günther for Rock and Roll, Dr.-Ing. Carsten Mehler and Dr.-
Ing. Michael Löffelmann for enjoying life in Munich. Especially I would like to thank
Carsten Sievers and Manuel Stratmann from the Lehrstuhl für Technische Chemie 2
at the Technische Universität München and Andreas Möller from the Institut für
Nichtklassische Chemie e.V. at the Universität Leipzig for investing their limited time
in adsorption experiments.
I thank all students and staff at the Lehrstuhl für Feststoff- und
Grenzflächenverfahrenstechnik whose presence and spirit made the otherwise
gruelling experience tolerable. Especially, I would like to thank Wolfgang Lützenburg
and Ralf-Georg Hübner for soccer.
I thank the company Grace Davison for providing zeolite material.
Last but not least I am grateful to my wife Özlem for the inspiration and moral
support she provided throughout my research work and for her patience. Without her
loving support and understanding I would never have completed my present work.
IV
Apollo and Daphne (1622-1625) from Gian Lorenzo Bernini
For my daughter Ilayda Defne Ingolstadt, 2006, Ugur Akgün
V
Table of Contents
Notation and Abbreviations VIII
1. Introduction 1
2. Objective 3
3. Fundamentals of Adsorption 6 3.1. Introduction 6
3.2. Adsorbents 6
3.2.1. Activated carbon 7
3.2.2. Zeolites 10
3.2.3. Silicalite-1 13
3.3. Characterising microporous solids 13
3.3.1. Specific surface 14
3.3.2. Pore volume distribution 16
3.3.3. Density and porosity 22
3.3.4. Chemical composition and structure 24
3.3.5. Refractive index 26
3.3.5.1. The Clausius-Mosotti equation 29
3.3.5.2. Refractive index of zeolites 33
3.3.6. Hamaker constant 39
3.3.6.1. Macroscopic approach of Lifshitz 40
3.3.6.2. Microscopic approach of Hamaker - de Boer 45
3.4. Adsorption isotherms 48
3.4.1. Types of isotherms 48
3.4.2. Langmuir isotherm 50
3.4.3. BET isotherm 52
3.4.4. Toth isotherm 55
3.4.5. Potential theories 56
VI
3.4.5.1. Polanyi and Dubinin 57
3.4.5.2. Monte Carlo Simulations 59
3.4.6. Isotherms based on the Hamaker energy of the solid 61
4. Experimental Methods for Adsorption Measurements 63 4.1. Volumetric method 63 4.2. Packed bed method 65 4.3. Zero length column method 67
4.4. Concentration pulse chromatography method 70
5. Prediction of Isotherms for p → 0 (Henry) 72 5.1. Thermodynamic consistency 74 5.2. Energetic homogeneous adsorbents 77
5.2.1. The dimensionless Henry curve 79
5.2.2. The van-der-Waals diameter of solids 85
5.3. Energetic heterogeneous adsorbents 87
5.3.1 The range of polarities 87
5.3.2. Induced dipole gas � electrical charge solid interaction (indcha)
87
5.3.3. Permanent dipole gas � induced dipole solid interaction (dipind)
90
5.3.4. Permanent dipole gas � electrical charge solid interaction
(dipcha) 91
5.3.5. Quadrupole moment gas - induced dipole solid interaction
(quadind) 92
5.3.6. Quadrupole moment gas - electrical charge solid interaction
(quadcha) 93
5.3.7. Overall adsorption potential 94
5.3.8. Henry coefficient for energetic heterogeneous adsorbents 97
5.4 Calculation methods for the Henry coefficient 98
5.4.1. Henry coefficients from experiments and literature 98
5.4.2. Theoretical Henry coefficients 99
VII
6. Prediction of Isotherms for p → ∞ (Saturation) 100 6.1. Differential form 101
6.2. Integrated form with boundary conditions 106
6.2.1. The dilute region 106
6.2.2. The saturation region 107
6.2.3. The general case 108
7. Experimental Results and Discussion 112 7.1. New experimental results 112
7.1.1. Adsorption isotherms 112
7.1.2. Comparison of Langmuir and Toth regression quality 114
7.1.3. BET surface and micropore volume 116
7.2. Prediction of isotherms 117
7.2.1. Henry coefficient predictions 117
7.2.2. Isotherm predictions 125
7.3. Model verifications 135
7.3.1. Refraction index verification 135
7.3.2. Dependency on the saturation degree 139
7.3.3. The London potential 142
7.3.4. Potential energy comparison and formula overview 144
8. Summary 149
Appendix 151 Appendix A: Isotherms in transcendental form 151
Appendix B: Materials and methods 155
B.1. Materials 155
B.1.1. Experimental data 155
B.1.2. Literature data 158 B.2. Experimental set-up and material data 159
B.2.1. Adsorption isotherm measurements 159
B.2.2. BET surface and micropore volume 162
B.2.3. Adsorptive data 163
VIII
B.2.4. Adsorbent data 164
Appendix C: Alternative way for deriving the potential factor fquadcha 165
References 167
IX
Notation and Abbreviations Latin Symbols
Ai,j - dimensionless interaction parameters
jSiAl
- ratio of aluminium to silicon atoms in the zeolite molecule
C - parameter in isotherm prediction equation
C - absorption strength in the IR and UV range
C J*m6 interaction constant
C mol/ m3 overall concentration of gas
D m diameter of a particle sphere
Dax m2/s axial dispersion coefficient in a pipe
E V/m field intensity
E J potential energy between two molecules or particles
F N force
HAds J/mol specific enthalpy of adsorption
HL J/mol specific heat of liquefaction
He mol/(kg*Pa) Henry coefficient
Ha J Hamaker constant
I A electrical current
I J ionisation energy of a molecule
IP - interaction parameter
K - effective isotherm slope in concentration pulse method
KL 1/Pa Langmuir constant
KT (1/Pa)m Toth constant
X
L m length of a pipe or pore capillary
M kg/mol molar mass
N mol number of moles
N& mol/s molar flow
N 1/m3 number density of dipole moments in a volume element
N - partial loading in Langmuir isotherm
NA 1/mol Avogadro constant = 6,0225*1023 1/mol
P C/m2 electrical polarisation
Q C*m2 quadrupole moment
R - parameter defined by equation (6.10)
R´ - parameter defined by equation (8.14)
ℜ J/(K*mol) universal gas constant = 8,3143 J/(K*mol)
Rm m3/mol molar refraction
Rm,i m3/mol molar refraction of the atom or ion
S m2 surface
SBET m2/kg specific surface of the adsorbent
T K temperature
V m3 volume
V m3 total volume of gas adsorbed on a solid surface
V& m3/s volumetric flow rate
W J work
a m/s2 acceleration
a - integration parameter
amol m2/number average area occupied by a molecule of adsorbate in the
BET monolayer
XI
ai mol/(m2*s*Pa) BET constant
a - slope
b - electronic overlap parameter
b kg/mol intercept of the linear form of the BET isotherm
bi mol/(m2*s) BET constant
b - interception value
c0 m/s speed of light in vacuum= 299792456.2 m/s
c - BET constant
c J/(kg*K) heat capacity
c mol/m3 gas concentration
d m tube inner diameter or pore diameter
d m distance (e.g. between particles)
e C charge of one electron = 1,60*10-19 C
f Hz oscillator strength
fi,j - potential factor
g Pa BET constant
h J*s Planck�s constant = 6,6256*10-34 J*s
i - complex number 1−
k J/K Boltzmann constant = 1,3804*10-23 J/K
l m distance between positive and negative charges
m kg mass
m kg/mol slope of the linear form of the BET isotherm
m - Toth parameter, 0 < m < 1
me kg mass of one electron = 9,11*10-31 kg
n - refractive index
XII
n - mean refractive index
n mol/kg loading
na - number of atoms in a formal solid molecule
nc - number of positive charges in a formal solid molecule
p Pa pressure
p0 Pa saturation pressure
p1 - parameter of the analytical Henry function
p2 - exponential parameter of the analytical Henry function
q C charge
r m distance
r m pore radius
s - average number of bonding electrons per solid atom
scations - average number of positive charges per solid atom
ss - sum of square error of the method of Margules
t s time
t nm statistical thickness of the adsorbed multilayer in the t-plot
u - substitution parameter for the Lambert function
v - number of valence electrons of the cation in the
stoichiometric formula of a zeolite
v m3/kg specific volume
vb m3/mol molar volume of adsorptive at normal boiling point
vmicro m3/kg micropore volume
vN2 m3/kg micropore volume measured with nitrogen adsorption
vvdW m3/mol van-der-Waals volume of a gas molecule
w m/s velocity
XIII
x - reduced distance
x1 - diameter ratio
xi - stoichiometric coefficient of matter i
y mol/mol molar fraction
y1 - diameter ratio
z m distance of the adsorptive from the adsorbent surface
Greek Symbols
Φ J potential of adsorption between an adsorptive molecule and
the solid continuum
Ψ − differential function of Toth
α (C2*m2)/J polarisability
α ′ m3 volumetric polarisability
α (C2*m2)/J average polarisability
( )Tβ m3/mol molar volume of adsorbate
β − affinity coefficient in Dubinin theory
β Grad contact angle
β J*m6 London constant
β J*m6 average London constant
χ − porosity
ε J/mol adsorption potential of Polanyi
ε(ω) − frequency dependent dielectric response function
ε0 C/(V*m) permittivity of free space = 8,854*10-12 C/(V*m)
εr - relative permittivity
XIV
ε C/(V*m) permittivity of a medium
ϕ Grad orientation angle of an adsorption pair to each other
ϕ − porosity
ϕ − ratio of partial pressure to saturation pressure
γ − material dependent constant
γ N/m surface tension
κ − function of the distribution of volume of the pores
λ W/(m*K) thermal conductivity
µ J chemical potential
µ C*m dipole moment (1 debye = 3,34*10-30 C*m)
µ0 (V*s)/(A*m) permeability of free space = 1.257*10-6 (V*s)/(A*m)
µr − relative permeability of a medium
µ (V*s)/(A*m) permeability of a medium
ν0 1/s frequency of the electron in the ground state
νe 1/s main electronic absorption frequency = 3*1015 1/s
νi 1/s orbiting frequency of the molecule electrons
π - circle number = 3.1416
ρ kg/m3 density
ρapp kg/m3 apparent density of the adsorbent
ρs kg/m3 true density of the adsorbent
ρj,at 1/m3 average number density of solid atoms per volume
σ C/m2 charge density
σ m van-der-Waals diameter
τ s mean retention time
τD s mean system dead time
XV
ω Hz frequency
ξ Hz imaginary frequency
Super- and Subscripts
A Avogadro
BET Brunauer, Emmet, Teller
C Carrier gas
G Gas
L Langmuir
L Liquid
L London
LG Liquid-Gas
S Solid
SG Solid-Gas
SL Solid-Liquid
at atom
app apparent
ads adsorption
adsads adsorbate � adsorbate
adsorb adsorbate
adsorp adsorptive
aft after
b bulk
bef before
XVI
c critical
calc calculated
dipcha permanent dipole - electrical charge
dipind permanent dipole � induced dipole
e electron
exp experimental
f free
i layer i
i atom or ion i
i adsorptive i
ij interaction of adsorptive i with adsorbent j
indcha induced dipole - electrical charge
indind induced dipole � induced dipole
j adsorbent j
jj interaction of adsorbent j with itself
m mole
max maximum
mic microporosity
micro micropore
mix mixed
mol molecule
mon monolayer
pred predicted
quadind permanent quadrupole � induced dipole
quadcha permanent quadrupole � electrical charge
XVII
r reduced
rel relative
ref reference
s saturation, solid
stand standard
vdW van-der-Waals
0 layer 0
0 vacuum
∞ infinite
* reduced
Abbreviations
BC before Christ
COSMO-RS conductor like screening model for real solutions
∆f free place
DFT density functional theory
G gas molecule
GCMC grand canonical Monte Carlo
IAST ideal adsorbed solution theory
IR infra red
IUPAC International Union of Pure and Applied Chemistry
LTA Linde type A zeolite
MD molecular dynamics
MPRAST multiphase predictive real adsorption theory
NP number of data points
OP occupied places
PBU primary building unit
SBU secondary building unit
XVIII
SP sorption place
UV ultra violett
ZLC zero length column
Dimensionless Numbers
Ads
C
Cj
jindind T
Tp
HaA 3
243
πσ= interaction parameter for the induced dipole � induced
dipole potential
( ) jiads
atjiindcha Tk
esA
,20
,22
8 σπερα
= interaction parameter for the induced dipole � electrical
charge potential
( ) 3,
20
,2
12 jiads
atjjidipind TkA
σπεραµ
= interaction parameter for the permanent dipole � induced
dipole potential
( ) jiads
atjidipcha Tk
esA
,22
0
,222
12 σπερµ
= interaction parameter for the permanent dipole � electrical
charge potential
( ) 5,
20
,2
403
jiads
atjjiquadind Tk
QA
σπερα
= interaction parameter for the quadrupole moment � induced
dipole potential
( ) 3,
220
,222
240 jiads
atjiquadcha kT
esQA
σπερ
= interaction parameter for the quadrupole moment �
electrical charge potential
23
3
=
cj
j
ads
c
pHa
TT
CBσ
parameter in the general isotherm
XIX
23
3
=′
cj
j
ads
c
pHa
TT
Bσ
parameter in the general isotherm
solidBETjmic SD ρσ= degree of microporosity
cj
jHa p
HaE 3
*
σ= reduced Hamaker energy
BETji STHeIL
,σℜ= initial loading
BET
Ai
SNn
N2σ
= number of molecule layers
solid
solidAsolid
MN
P0
*
ερα
= solid parameter
micro
iBET
vS
Sσ
=∗ microporosity parameter
c
adsadsr T
TT =_ reduced adsorption temperature
( ) ( )T3* βσβ iANT = reduced molar volume of adsorbate
pHenn =* reduced loading
( )microadsorbmicro
is v
Tnv
Mnn β
ρ==* saturation degree or pore filling
adsorbmicro
is v
Mnn
ρ−=− 11 * residual capacity
ij
zxσ
= reduced distance
1
1. Introduction The term adsorption denotes the mass transfer of substances from a fluid phase
towards a solid surface. Thus, adsorption is a surface effect on solid phases.
Accordingly one or more components (adsorptives) of a fluid (liquid or gas) are
attached as adsorbate onto solids (adsorbents). The opposite effect is called
desorption. Adsorption is a mass transfer mechanism in the two phase system
solid/fluid [47]. At a given temperature this mass transfer leads to a thermodynamic
equilibrium at which the concentrations of adsorptives and adsorbates remain
constant. In the dilute region the solid adsorbate loading is proportional to the partial
pressure of the adsorptive. The proportional constant is called Henry coefficient. At
higher pressures the loading achieves saturation.
The ability of porous solids to reversibly adsorb large volumes of vapour was
detected in the eighteenth century [48]: Lowitz studied in 1785 the effectiveness of
charcoal in decolorizing various aqueous solutions and, in particular, its commercial
application to the production of tartaric acid. De Sussure reported in 1812 on
adsorption experiments of different gases onto porous substances whereas he also
described the heat of adsorption. In 1876 Gibbs thermodynamically derived the law of
adsorption by describing the relation of surface loading with gas or liquid
concentration. Advances towards a quantitative description of the adsorption
mechanisms were made in the twentieth century. Besides Freundlich and Toth,
Langmuir, in particular, has to be mentioned. In 1916 he derived the theory of
monomolecular covering of adsorbed surfaces. This concept was expanded to the
theory of multilayer adsorption by Brunauer, Emmet and Teller (BET) in 1938. In
1914 Polanyi introduced the adsorption potential to describe the dependency of the
adsorption equilibrium on temperature [24]. Dubinin extended this potential theory in
1959.
The laws of equilibrium thermodynamics, firstly applied to adsorption by Prausnitz in
the nineteen sixties, were adopted to gas mixtures by Myers. He developed the
theory of ideal adsorbed solutions (IAST) [24]. The developments are still ongoing. A
2
good example is the multiphase predictive real adsorption theory (MPRAST), by
which Markmann described the equilibrium of a non-ideal adsorbate mixture with an
energetically non-homogenous solid surface [8].
Some modern concepts have to be mentioned also. With the Hellman-Feynman-
Theorem it is possible to specify intermolecular interactions. The basic idea is that
the interaction correctly and completely can be described with the help of
electrodynamic approaches if the charge distribution of the molecules is known [49].
The charge distribution itself can be calculated by applying the density functional
theory as in the COSMO-RS-approach.
Mehler [50] found that it is still not possible to describe real solid surfaces using only
theoretical quantum mechanical calculations. Surfaces seem to be too complex and
unknown. On the contrary, he used adsorptive molecules with known properties as
probes to characterize the solid surface. The interaction of these probes with the
adsorbent is the basis in COSMO-RS for calculating the energy distribution of the
solids. He quantified the interaction potentials of the solids via the Henry-coefficients
determined by adsorption experiments.
It seems that Henry-coefficients are very important not only for describing the
adsorption itself but also for characterizing the solid surfaces. The basis of all
adsorption processes and theories is the single-component adsorption equilibrium
and especially the Henry-coefficient. Maurer established an innovative way for a
priori calculating Henry-coefficients, based only on gas and solid properties. He
analysed very successfully the interaction of gases with energetically homogenous
microporous adsorbents like activated carbon.
Nevertheless, there is no general model for the description of adsorption of any polar
or nonpolar gases on any polar or nonpolar microporous adsorbent. After removing
this deficiency it would be possible to improve the prediction not only of
multicomponent adsorption problems which are real engineering tasks. Also the
characterisation of solid surfaces with density functional theories would be one
important step closer.
3
2. Objective Hydrogen plays an eminent role in the processing of fuels and can become important
for engines in motorcars with respect to CO2-emissions. An important task is the
design of a new unit for the purification of a contaminated hydrogen gas in a
petrochemical plant. This gas is used in heterogeneous hydrogenation reactors with
fixed bed catalysts. The separation of the contaminants from hydrogen leads to a
very pure hydrogen feed to the reactors resulting in an improved selectivity. The tail
gas of the separation no longer burdens the capacity of the whole petrochemical
plant and can be burned as recycled fuel gas in the cracking furnaces. A feasible
process is a pressure swing adsorption unit to purify the hydrogen stream. When
planning this new separation process an engineer has to consider the two main
general aspects
- phase equilibrium and
- mass transfer dynamics.
Understanding the mass transfer dynamics aids the calculation of the time required
for the process. In the above example it is the time until the adsorption beds are on
the verge of a breakthrough of contaminated hydrogen feed. For continuous
operation it is necessary to switch the feed from this loaded adsorber to a second
one which is fresh regenerated. In modern concepts automated programs control the
switching matrix of the adsorber valves. The dynamics of the process yields the
information when to switch the valves so that the whole capacity of the adsorbers can
be used without letting contaminated gas through.
The first aspect is the equilibrium of the materials in the separation process itself. It
gives the engineer information about the amount of adsorbent required to remove the
impurities in hydrogen stream and to calculate the size of the apparatus. In
combination with the dynamics of the process two general engineering challenges
can be optimised: Maintaining a constant high quality of the product and reducing
expenses.
4
In adsorption processes or adsorption steps of physical, chemical or biological
reactions the adsorption isotherm is the most used description method of the phase
equilibrium. Knowledge of phase equilibria is the key for understanding complicated
processes. Engineers need these isotherms in the planning of production processes.
On the other hand it consumes both time and manpower to determine the isotherms
by costly experiments.
Besides engineering aspects the scientific viewpoints of adsorption are of great
interest when developing new materials and processes. The physical behaviour of
complex systems of gases and solids often depends on the adsorption of the educts.
For instance, in heterogeneous catalysis gas molecules adsorb on the catalyst
surface, and reactions in the adsorbed monolayer lead to products which are
desorbed from the surface.
Many processes and reactions in the chemical industry and in chemical science are
based upon heterogeneous catalysis. Three important questions of these processes
are the understanding of the reaction, the development of the catalyst and the design
and optimisation of the reactor. All aspects can benefit from a rigorous and accurate
kinetic modell. Especially the competition for adsorption sites is very important for the
kinetics of a heterogeneous catalytic reaction if it is the rate determining step. It refers
then to the slowest step of the reaction process and controls the activity [123]. A
prerequisite for kinetic modelling is the knowledge of adsorption equilibria.
Therefore, the adsorption equilibrium of the gaseous educts with the catalyst surface
is related to the reaction rate and also the amount of molecules occupying the free
adsorption sites. In both main model mechanisms of heterogeneous catalysis
(Langmuir-Hinshelwood and Elay-Rideal) the adsorption isotherm is needed to
calculate the effective reaction rate in dependence of the partial pressures of the
reactands. Therefore adsorption equilibria are not only eminent for the understanding
of the process of heterogeneous catalysis but also for the development of novel
catalysts and adsorbents.
5
In the literature the isotherm of interest is not always available. Concepts like
COSMO-RS are very promising but to complex for daily engineering work. Also
adaptations on the isotherm data like Langmuir and Toth can not always be
extrapolated to other gas-solid systems. There is no general model to describe the
adsorption of any polar or nonpolar gas by any polar or nonpolar microporous
adsorbent.
The objecitve of this study is driven by this demand. First, the fundamentals of
adsorption will be examined in this work. The engineer and scientist will get an
overview over microporous adsorbents, methods of measurement and the different
isotherm descriptions. There will be a Section about characterisation of solid
surfaces. Because there is a lack of data regarding Hamaker constants of
microporous solids an advanced Hamaker-de Boer model will be introduced to
calculate this dispersion coefficient.
Henry coefficients of adsorption will a priori be calculated based on gas and solid
properties of the molecules. In the case of nonpolar gas � nonpolar solid, polar gas �
nonpolar solid and nonpolar gas � polar solid an analytical equation will be
introduced to directly calculate the Henry coefficients. This approach will be
expanded to the fourth case of polar gas � polar solid so that the whole spectrum of
possibilities is covered. A novel model for predicting the complete adsorption
isotherm based only on the molecular properties will be introduced. The model will be
verified with own experimental and literature data.
After a discussion of the experimental and theoretical results there will be a
conclusion and an outlook to future demands. The experimental methods and
materials of the work will be explained in the appendix together with lists of the
properties of the examined adsorptives and adsorbents.
6
3. Fundamentals of Adsorption 3.1. Introduction Although adsorption has been known as a physical process since the 18th century
new aspects are generated through the demands of modern engineering and science
in the 21st century.
On the one hand the industry, especially the chemical one, has to improve itself
permanently. The understanding of adsorption is a key for future improvements. To
withstand the global competition the industry has to enhance its effectiveness
continuously. Manpower and invests need to be adopted very carefully at the right
time. So modern engineers need powerful instruments to base their decisions on
solid technical fundamentals. These instruments need to be economic, usable in a
wide range of applications and easy to learn. This study provides such an instrument
for the field of adsorption.
On the other hand science also has made great advances and still continues to.
Adsorption seems to be a fundamental physical step in a wide range of research
fields. For example in medical biochemistry the interaction of cells with biomaterials is
dependent on the adsorption layer of proteins [51]. Another example is the storage of
hydrogen through reversible adsorption in metallic hydrides. Also the adsorption step
of reactants on solid surfaces is eminent for the field of heterogeneous catalysis.
The aim of Chapter 3 is to give a brief overview on the state of research and the
literature regarding adsorption science and processes.
3.2. Adsorbents
Development and application of adsorption cannot be considered separately from
development of the technology of manufacture of adsorbents applied both on the
laboratory and industrial scales. These sorbents can take a broad range of chemical
7
forms and different geometrical surface structures. This is reflected in the range of
their applications in industry, or helpfulness in the laboratory practice [52]. Table 3.1
shows the basic types of industrial adsorbents.
Table 3.1: Basic types of industrial adsorbents.
Carbon adsorbents Mineral adsorbents Other adsorbents
Activated carbons Silica gels Synthetic polymers
Avtivated carbon fibres Activated alumina Composite adsorbents: (complex mineral_carbons, X_elutrilithe, X_Zn, Ca.)
Molecular carbon sieves Oxides of metals Mixed sorbents
Mesocarbon microbeads Hydroxides of metals
Fullerenes Zeolites
Heterofullerenes Clay minerals
Carbonaceous nanomaterials
Pillared clays
Porous clay hetero-structures (PCHs)
Inorganic nanomaterials
In this work emphasis is laid on microporous adsorbents like activated carbon and
zeolites, especially aluminosilicates. These sorbents play an important role in industry.
3.2.1. Activated carbon
Since our early history active carbon has been the first widely used adsorbent. Its
application in the form of carbonised wood (charcoal) was described as early as 3750
BC in an ancient Egyptian papyrus [52].
Nowadays commercial sources for activated carbons include biomass materials (wood,
coconut shell and fruit pits) and fossilized plant matter (peats, lignites and all ranks of
coal) or synthetic polymers [53]. Activated carbon is normally made by thermal
8
decomposition of carbonaceous material followed by activation with steam or carbon
dioxide at elevated temperature (700 � 1100°C). The activation process essentially
involves the removal of tarry carbonisation products formed during the pyrolysis,
thereby opening the pores [48]. It is generally believed that carbon dioxide activation
mainly causes the creation of microporosity while steam activation results in the
development of mesoporosity to a higher extent. These types of activation are known
as the physical methods and are widely used in industries.
Important parameters that determine the quality and yield of activated carbons are the
final temperature of activation and dwell time at this temperature [54]. Effects of dwell
time and activation temperature on the BET surface area and micropore volume of
activated carbons made from pistachio-nut shells are shown in figure 3.1 and in figure
3.2 respectively. Also the CO2 flow rate and the heat rate have an impact.
The structure of activated carbon consists of elementary micro crystallites of graphite,
but these micro crystallites are stacked together in random orientation. The spaces
between the crystals which form the micropores. The pore size distribution is typically
trimodal [48]. Characteristic properties of activated carbons used for gas purification
are listed in table 3.2 [24].
Figure 3.1: Effects of dwell time on the BET surface area and micropore volume of
activated carbons made from pistachio-nut shells [54].
9
Figure 3.2: Effects of activation temperature on the BET surface area and micropore
volume of activated carbons made from pistachio-nut shells [54].
Table 3.2: Data of different activated carbons for gas purification [24].
Characteristic properties Numbers Units
True density ρ 2000 kg/m3
Apparent density ρapp 600 - 800 kg/m3
Bulk density ρb 350 - 500 kg/m3
Porosity ϕ 0,7 - 0,6 -
Specific surface SBET 900.000 - 1.200.000 m2/kg
Macropore volume d > 50 nm 0,3*10-3 - 0,5*10-3 m3/kg
Mesopore volume 2 nm < d <
50 nm
0,05*10-3 � 0,1*10-3 m3/kg
Micropore volume d < 2nm 0,25*10-3 � 0,5*10-3 m3/kg
Specific heat capacity c 840 J/(kg*K)
Thermal conductivity λ 0,65 W/(m*K)
10
3.2.2. Zeolites
The main class of microporous adsorbents are zeolites which comprise, according to a
recommendation of Structure Commission of the International Zeolite Association, not
only aluminosilicates but also all other interrupted frameworks of zeolite-like materials,
e.g. aluminophosphates [52]. In this study emphasis is laid on aluminosilicates like
MS4A, MS5A, NaX and NaY which are widely used in industry.
The term zeolite was originally coined in the 18th century by a Swedish mineralogist
named Axel Fredrik Cronstedt who observed, upon rapidly heating a natural mineral,
that the stones began to dance about as the water evaporated. Using the Greek words
which mean "stone that boils" he called this material zeolite.
The elementary building units of zeolites are SiO4 and AlO4- tetrahedrons. Adjacent
tetrahedrons are linked at their corners via a common oxygen atom, and this results
in an inorganic macromolecule with a structurally distinct three-dimensional
framework. It is evident from this building principle that the net formulae of the
tetrahedrons are SiO2 and AlO2, i.e. one negative charge resides at each tetrahedron
in the framework which has aluminium in its centre. The framework of a zeolite
contains channels, channel intersections and/or cages with dimensions from ca. 0.2
to 1 nm. Inside these voids are water molecules and small cations which compensate
the negative framework charge [55].
Figure 3.3 shows the principle construction of a zeolite. The smallest unit is the
primary building unit (PBU) which consists of SiO4 or AlO4- tetrahedrons. These
tetrahedrons are linked together via oxygen ions to secondary building units (SBU).
The SBUs can be of cubic form or a hexagonal prism or an octahedron. They are
knotted to β-cages which are linked again via the oxygen ions [56].
The chemical composition of a zeolite can, hence, be represented by a formula of the
type Mx/v((AlO2)x(SiO2)y)z*H2O [57] with v the number of valence electrons of the cation
M (Na+, K+, Ca++, Mg++). The numbers x and y are integers and the ratio y/x is equal or
bigger one, z is the number of water molecules in the zeolite unit cell [57].
11
Figure 3.3: Principle construction of a zeolite of LTA type [56].
Si4+O2-
Si Si
12
Until today approximately 40 natural zeolites have been found. As the consequence
of successful synthesis work more than 100 new zeolites were produced. Most of the
zeolites are manufactured via so-called hydrothermal synthesis. These are reactions
in aqueous solution with temperatures above 100°C, similar to the natural conditions.
In the laboratory the sources for the reactions are sodium silicate or silica sol as the
SiO2-origin and hydrated aluminium hydroxide or aluminates as the Al2O3-origin. The
two sources react with strong bases like NaOH or KOH under hydrothermal
conditions and pressure in autoclaves. The reaction time can be between several
minutes and months [58].
Zeolites are applied in many fields.
- One characteristic property is the ability of ion exchange. Zeolite A can
exchange its sodium ions with the calcium ions in water which becomes
then �softer�. The calcium enriched zeolites are discharged with the waste
water.
- Zeolites can act as acid catalysts in chemical reactions because the alkali
metal ions can be exchanged with protons. This happens in the inner
cavities, so zeolites can be handled without danger.
- Another application is the use of zeolites as chemical sensors for detecting
gases like NH3 in exhausts of automobiles or power plants. It seems that
protons can migrate through the zeolite cavities. This causes a change in
the ionic conductibility which is measurable.
- For this work the adsorption of gases or vapours is eminent. As adsorbents
they can capture air moisture or solvent vapour. They are used in the
clearance of multiple composite glasses or for drying of refrigerants in
cooling circuits or brake systems.
- Sterical effects of zeolites are neglected in this study. However they can be
very important in many chemical processes like filtering linear
hydrocarbons out of petrol to get it knockproof and, in general, separation
of mixtures.
13
3.2.3. Silicalite-1
Silicalite-1 was firstly introduced by Flanigen as a new hydrophobic but organophilic
polymorph of silica. This new polymorph was topologically very similar to ZSM-5 so it
was considered as the aluminium free end member of MFI type zeolites. It contains
two types of channel systems with similar size: straight channels and sinuous
channels. The diameters of these channels are about 0.54 nm. These two different
channels are perpendicular to each other and generate intersection areas which
have 0,89 nm of diameter. Silicalite-1 has ten membered oxygen pore openings, and
has high thermal and acidic stability. The comparable pore size of silicalite-1
channels with industrially important small organic molecules like methane, n-butane,
n-octane and ethanol makes it important for many applications [59]. Silicalite-1
provides a non-polar structure [133]. It is, therefore, considered as a model adsorbent
for testing adsorption isotherm models, based on gas-solid molecular interactions
[60].
3.3. Characterising microporous solids Most analysis techniques typically probe a particular aspect of the material and,
consequently, a combination of methods is necessary to give a balanced description
of complex solids [61]. Microporous adsorbents have very different properties which
are fundamental for the process of adsorption. Their properties can influence the
equilibrium and/or the dynamics of adsorption. Although this study is mainly about
gas adsorption equilibria also dynamical aspects should be mentioned because
engineers need both information for planning industrial or laboratory processes. The
dynamic and steady state properties of the solids can be found by characterising
them through following factors:
- Specific surface
- Porosity
- Density
- Chemical composition
- Refractive index
- Hamaker constant
14
3.3.1. Specific surface
There are interactions of molecules on every interface between a solid and a fluid �
gas or liquid- phase which can lead to attraction or repulsion forces. In the case of
attraction a two dimensional interaction phase is composed with the thickness of
molecule dimensions. This phenomenon is usable for technical purposes if this
interface has sufficient dimensions and if it is accessible for the molecules. The
surfaces of microporous adsorbent have magnitudes of several hundred square
meter per gram solid [24].
From type II and type IV isotherms, the specific surface area of the adsorbents can
be determined by applying the BET method [61]. Brunauer, Emmett and Teller have
derived an equation for the adsorption of gases on solids with n layers [62]:
( )( )( ) 1
1
, 1111
1 +
+
−−+++−⋅
−= n
nn
monG
G
ccnnc
VV
ϕϕϕϕ
ϕϕ (3.1)
where VG is the total volume of gas adsorbed on the solid surface, VG,mon the volume
of gas adsorbed in a monolayer, ϕ the ratio of pressure to saturation pressure of the
gas. The parameter c is a dimensionless constant and dependent from the enthalpy
of adsorption in the first layer and the heat of evaporization. The assumption made by
Brunauer, Emmet and Teller is that the evaporation-condensation properties of
molecules in the second and higher adsorbed layers are the same as those of the
liquid state.
If we deal with adsorption on a free surface, then at the saturation pressure of the
gas an infinite number of layers can build up on the adsorbent. ∞→n one leads to
( ) ( )ϕϕϕ
111
1, −+⋅
−=
cc
VV
monG
G (3.2)
15
With the law of ideal gases and the balance of the mass of the adsorbent equation
(3.2) becomes
( ) ( )ϕϕϕ
111
1 −+⋅
−=
cc
nn
mon
(3.3)
where n is the overall loading of the adsorbent and nmon the monolayer loading. From
this the linear form (y = b + m ϕ ) can be derived to obtain the specific surface of the
adsorbent due to the BET adsorption isotherm:
( )( )ϕ
ϕϕ
cnc
cnn monmon
111
−+=−
(3.4)
The intercept, b =cnmon
1 , and the tangent of the gradient angle, m = ( )cn
c
mon
1− , permit the
calculation of the monolayer loading nmon and the BET constant c:
mbnmon +
= 1 (3.5)
1+=bmc (3.6)
Finally the BET surface area is calculated from
AmolmonBET NanS = (3.7)
Here amol denotes the average area occupied by a molecule of adsorbate in the
completed monolayer and NA the Avogadro constant [63].
By convention, most surface area determinations are based on the area occupied by
a nitrogen molecule. The nitrogen adsorption is measured at the boiling point of
nitrogen, 77.4 K. The value of amol may be estimated following an early suggestion of
Emmett and Brunauer from the density of liquid nitrogen at 77.4 K. The estimate uses
16
the assumption that the packing density of adsorbed nitrogen on a surface is the
same as in the liquid. This leads to the equation
32
091,1
=
ALmol N
Maρ
(3.8)
where M is the molar mass of nitrogen (0.028 kg/mol), Lρ the density of liquid
nitrogen (810 kg/m3) and 1.091 is a packing factor for 12 nearest neighbours in the
bulk liquid and six on the plane surface. Insertion of these quantities and of the
Avogadro constant into equation (3.8) gives amol(N2) = 0.162 nm2 for nitrogen at 77.4
K.
For practical purposes, the average area occupied by other molecules can be
determined by comparison of their monolayer capacity with that of nitrogen. Using
this method, the areas occupied by argon amol(Ar) and krypton amol(Kr) at 77.4 K were
then determined to be 0.138 nm2 and 0.2 nm2, respectively.
3.3.2. Pore volume distribution
The word pore comes from the Greek word �ποροσ� which means passage. This
indicates the role of a pore acting as a passage between the external and the internal
surfaces of a solid, allowing material, such as gases and vapours, to pass into, through
or out of the solid. Almost all adsorbents used in catalysis or for purification/separation
purposes have a high porosity, and this porosity is the only practical method of
introducing greatly enhanced surface areas into a solid [64]. Porosity has a
classification system as defined by IUPAC which gives a guideline of pore widths
applicable to all forms of porosity. Distinctions in porosity classes are not rigorous and
they may often overlap in size and definition. The widely accepted IUPAC classification
is as follows:
dMicropores < 2 nm
2 nm < dMesopores < 50 nm
dMacropores > 50 nm
17
There are several methods to measure pore related parameters of a solid. Here
exemplarily the mercury porosimetry and the t-plot method of de Boer shall be
discussed in detail. The other common methods are listed in Table 3.3 [67]:
Table 3.3: Methods to measure pore related parameters of a solid.
Method Definition
BJH The method of Barrett, Joyner and Halenda is a procedure for
calculating pore size distributions from experimental isotherms using
the Kelvin model of pore filling. It applies only to the mesopore and
small macropore size range.
de Boer t-Plot The t-Plot method is most commonly used to determine the external
surface area and micropore volume of microporous materials. It is
based on standard isotherms and thickness curves which describe
the statistical thickness of the film of adsorptive on a non-porous
reference surface.
MP-Method The MP method is an extension of the t-plot method. It extracts
micropore volume distribution information from the experimental
isotherm.
Dubinin Plots Dubinin plots (Dubinin-Radushkevich plot and the more general
Dubinin-Astakhov plot) relate the characteristic energy of the
adsorptive to the micropore structure.
Medek The Medek method uses Dubinin-Astakhov plots to determine
micropore volume distributions by pore size.
Horvath-
Kawazoe
technique
This method determines the best fit (in a least squares sense) of a
set of single-mode model isotherms to the experimental isotherm.
The solution set represents the pore volume distribution by size for
the solid on which the isotherm was developed.
DFT plus Density functional theory provides a method by which the total
expanse of the experimental isotherm can be analysed to determine
both microporosity and mesoporosity in a continuous distribution of
pore volume with respect to pore size.
18
In mercury porosimetry mercury is injected into the pores of the solid to measure the
pore size and the pore volume distribution. The gas is evacuated from the sample cell
and mercury is then transferred into the sample under vacuum and pressure is applied
to force mercury into the sample. In the theoretical model of Washburn [66] all pores
are treated as cylindrical pores with an inner radius r. The mercury with the pressure p
is balanced with the surface tension γ of mercury. This leads to
βγ cos2−=rp (3.9)
with β the contact angle of the mercury with the pore surface. A typical value for β is
140 degrees and for the surface tension 484 mN/m or 0.484 J/m2.
The Washburn equation (3.9) can be derived from the equation of Yang and Dupré
[65]:
βγγγ cosLGSLSG += (3.10)
where γSG is the interfacial tension between solid and gas, γSL the interfacial tension
between solid and liquid, γLG the interfacial tension between liquid and gas and β the
contact angle of the liquid on the pore wall. It is derived from a force balance as seen
in figure 3.4 [49].
Figure 3.4: Force balance of a mercury drop on a solid surface.
The work, W, required for moving the liquid up the capillary when the solid-gas
interface disappears and solid-liquid interface appears is
γSG
γLG
γSL
solid (S)
liquid (L)β
19
( ) SW SGSL ∆−= γγ (3.11)
where ∆S is the surface of the capillary wall covered by liquid when its level rises.
With equation (3.10) it is
SW LG ∆−= βγ cos (3.12)
The work required to raise a column of liquid in a capillary with the radius r is identical
to the work necessary to force liquid out of the capillary. If a volume V of liquid is
forced out of the capillary with a gas at a constant pressure p, the work W is
pVW = (3.13)
A combination of equations (3.12) and (3.13) gives
SpV LG ∆−= βγ cos (3.14)
When the capillary is circular in cross-section, parameters V and ∆S are given by
Lr 2π and rLπ2 , if L is the length of the capillary. Substituting these terms into
equation (3.14) yields the Washburn equation (3.9), which relates the pressure of the
mercury to the radius of the pores via an indirect proportionality.
The interrelationship of the differential volume dV with the differential radius dr is
given with a pore volume distribution function f(r)
drrfdV )(= (3.15)
The Washburn equation (3.9) can be rewritten in
dpprdr −= (3.16)
The combination of the equation (3.15) with (3.16) yields
20
−=
dpdV
rprf )( (3.17)
With equation (3.17) the pore radius r is related to the pore volume distribution f(r).
Another illustration can be found with the relationship
drr
dpp −= (3.18)
combined with
rddVdV
drddV
dpprfr
ln)( ==−= (3.19)
Therefore, the distribution function f(r) and the related volume distribution dV/dlnr can
be calculated.
Another method for measuring pore volumes is the t-plot method [63]. The t-plot
method is most commonly used to determine the external surface area and
micropore volume of microporous materials. It is based on standard isotherms and
thickness curves which describe the statistical thickness of the film of adsorptive on a
non-porous reference surface.
For a large number of non-porous solids the shape of the nitrogen isotherm can be
represented with a single curve, by plotting the ratio SVN2 as a function of the relative
pressure 0pp .
2NV denotes the volume of N2 adsorbed, S the surface area of the
sample, p and p0 the partial and saturation pressure of nitrogen, respectively. The
resulting t-curve reflects the statistical thickness t of the adsorbed nitrogen multilayer
and can be calculated according to:
21
( )
+
=
pp
nmt0
log034.0
99.131.0 (3.20)
Deviations of the t-plot from the ideal behaviour allow the deduction of the nature of
pores and the determination of the micropore volume (figure 3.5). Curve a in figure
3.5 is linear and starts at the origin, which indicates an ideal t-material. Therefore, S
can be calculated from the tangent. Curve b shows a strong upward deviation from
linearity at a certain pressure 0pp , which indicates that starting from t1 capillary
condensation takes places in addition to adsorption. In contrast, if some narrow pores
are filled by multilayer adsorption according to curve c further adsorption does not
occur in this part, as that surface has become unavailable. At this point the t-plot
begins to deviate downwards from the straight line at t2. This situation is, for instance,
encountered in the presence of slit-shaped pores. The presence of micropores is
indicated by a positive intercept at the y-axis as shown in line d. A straight line is
found and the surface S can be calculated from this tangent. The positive intercept
point is caused by a relatively large nitrogen uptake at very low t-values. Such a high
nitrogen uptake at very low pressures is due to the strong adsorption in micropores
and, hence, allows to calculate the micropore volume vmicro from the intercept.
Figure 3.5: Different courses of the t-plot.
t(nm)
vN2(m3/kg)
a
c
b d
vmicro
t2 t1
22
For more detailed descriptions and theories to porosity and measurement methods
reference should be made to the literature [68].
3.3.3. Density and porosity
The density of a porous solid is the relation of its mass mS to its volume VS without
the pore volume:
S
SS V
m=ρ (3.21)
Pycnometric density is at the moment the closest approximation of true density [70].
The term pycnometer is derived from the Greek word �πυκνοσ�, meaning dense, and
meter. With this method the pores are filled with gas molecules [69]. Helium is
recommended as probe gas [69] because:
- it is atomic and very close to ideal gas behaviour,
- it is not adsorbed at room temperature,
- its atoms are very small to diffuse also into smallest pores and
- it has a high heat transfer coefficient to adapt the chamber temperature
very quickly.
Figure 3.6: Draft of a gas pycnometer.
extra volumes
V3 big
V2 small sample
pressure
V1 Measurement chamber
v1 v2 v3
23
V1, V2 and V3 are the calibrated volumes of the measurement chamber and the extra
volumes for expansion of the measurement gas. Vgas is the volume of the gas in the
measurement chamber with the sample in it. After weighting the chamber without
(mchamber) and with the sample (mchamber+sample) the mass of the sample can be
calculated.
chambersamplechambersample mmm −= + (3.22)
In the first step the sampled chamber is filled with a well chosen gas and the
pressure p1 after equilibrium is measured. In the second step the valve v2 between
the measurement chamber and the small expansion chamber is opened followed by
a second pressure measurement p2. With the law of ideal gases it follows
( )221 VVpVp gasgas += (3.23)
or
12
1
2
−=
ppV
Vgas (3.24)
For the true density it follows:
12
1
21
−−
−= +
ppV
V
mm chambersamplechamberSρ (3.25)
In order to calculate the volume of the pores the helium in the measurement chamber
is moved by a pump. The chamber (with the solid in it) is filled again with mercury
under atmospheric conditions. The mercury encapsulates the solid without
penetrating into the pores. So it is possible to calculate the pore volume of the
sample:
24
mercurygaspores VVV −= (3.26)
Finally the porosity of the microporous adsorbent is
1
11
111
1
21
2
1
21
2
1
21
2
−
−
−
=−
−−
=−+−
−−
=+
=
mercury
mercury
mercury
mercury
gasmercurygas
mercury
spores
pores
VV
ppVV
VV
VppV
VVVV
VppV
VVV
ϕ
(3.27)
The apparent density of the microporous solid is
mercury
chambersamplechamber
spores
chambersamplechamberapp VV
mmVVmm
−−
=+−
= ++
1
ρ (3.28)
and defined as the ratio of mass of the solid adsorbent to the volume of the solid
inclusive the pore volume.
3.3.4. Chemical composition and structure
The chemical composition and the structure of a solid play an important role for
polarity and adsorption behaviour similar to the density and the porosity. Zeolites are
the main interest of this study so emphasis is laid on composition of aluminosilicates.
The structures of zeolites consist of three-dimensional frameworks of SiO4 and AlO4-
tetrahedron. The aluminium ion is small enough to occupy the position in the centre
of the tetrahedron of four oxygen atoms, and the isomorphous replacement of Si4+ by
Al3+ produces a negative charge in the lattice. The net negative charge is balanced
by exchangeable cations (sodium, potassium, or calcium). The ratio of silicon to
aluminium ions cannot be less than one and is infinite in silicalite-1. Zeolite synthesis
is already explained in Section 3.2.2. The size of the cage window is determined by
the number of oxygen atoms in the ring. A consequence is the physical effect of
sieving molecules out of a gas mixture. In table 3.4 important zeolites are listed [24].
25
Table 3.4: Chemical composition of technical important molecular sieves.
Structure Cations Trivial name Chemical composition Effective
pore opening
(10-10 m)
A Na+ MS4A Na12((AlO2)12(SiO2)12) 4.2
A Ca2+ MS5A Ca5Na2((AlO2)12(SiO2)12) 5.0
A K+ MS3A K12((AlO2)12(SiO2)12) 3.8
10X Ca2+ Ca40Na6((AlO2)86(SiO2)106) 8
13X Na+ NaX Na86((AlO2)86(SiO2)106) 9 to 10
Y Na+ NaY Na56((AlO2)56(SiO2)136) 9 to 10
ZSM-5 Na+ Na3((AlO2)3(SiO2)93) 6
In Table 3.5 the effect of sieving of zeolites regarding to different gas molecules is
listed. The critical molecule diameter in 10-10 m is in brackets [24].
Table 3.5: Sieving effect of zeolites.
Zeolite Effective
pore opening
in 10-10 m
Adsorbed molecules, critical diameter in 10-10 m
MS3A 3.8 He (2), Ne (3.2), Ar (3.83), NH3 (3.8), H2 (2.4), N2 (3.0), O2
(2.8), H2O (2.6)
MS4A 4.2 Kr (3.94), Xe (4.37), CH4 (4.0), C2H6 (4.44), C2H2 (2.4),
CH3OH (3.0), CO2 (2.8), H2S (3.6), every of above
molecules can be adsorbed
MS5A 5.0 CF4 (5.33), C2F6 (5.33), Cyclopropane (5.0), every of
above molecules can be adsorbed
10X 8 SF6 (6.7), i-C4H10 (5.6), C6H6 (6.7), Triophene (5.3), every
of above molecules can be adsorbed
13X
(NaX)
9 to 10 every of above molecules can be adsorbed
Y (NaY) 9 to 10 every of above molecules can be adsorbed
26
The molecule diameter can be calculated with the van-der-Waals volume vvdW (in
m3/mol) which is defined as [26]
c
cvdW p
Tv
8ℜ
= (3.29)
where c relates to critical properties of temperature and pressure and ℜ is the
universal gas constant. With the assumption that the gas molecule is a sphere the
molecule diameter can be calculated. With
3
234
== dV
Nv
sphereA
vdW π (3.30)
it follows for the molecule diameter in m that
316
3π
σAc
ci Np
Tℜ= (3.31)
With equation (3.29) and table 3.5 it can be decided whether or not a zeolite is suited
for a technical sieving process.
3.3.5. Refractive index
It will be shown later that the refractive index n and the interacting energies Φi,j
between adsorptive and adsorbent molecules are key parameters in order to
calculate adsorption equilibria. Adsorptive molecules can be nonpolar or polar with a
dipole moment µi and/or a quadrupole moment Qi. In polar molecules the centres of
the positive and negative charge q have the distance l:
lqii =µ (3.32a)
If two dipoles with counterdirection are effective in an adsorptive molecule a
quadrupole Qi can be effective.
27
With respect to zeolite molecules the anions and cations lead to an electrical field
with the strength E according to
204 rqE
πε= (3.32b)
with ε0 as the electrical permittivity in vacuum and r is the distance between the
charges. In electrical fields a dipole moment µi,ind can be induced in an adsorptive
molecule according to
( ){
EkT
E
polar
i
unpolariiindi
44 344 21
+==
3
2
0,µααµ (3.32c)
The polarisability (α0)i of nonpolar or αi of polar molecules are definded by this
equation. The interaction energies Φi,ind according to induced dipoles, ( )iµΦ
according to permanent dipoles µi,ind and ( )iQΦ due to a quadrupole moment Qi are
( )
( )rEQ
Q
andE
E
ii
ii
iindi
∂∂−=Φ
−=Φ
−=Φ
2
22
,
µµ
α
(3.32d)
As mentioned above the refractive index is one of the important parameters
describing the optical properties of solid materials. In general it is difficult to obtain a
quantitative relation between the refractive index and the structure as well as
composition of materials. It seems that the refractive index is dependent on atomic
parameters like mass, radius and electric charge of the ions constituting the material
[71]. According to classical dielectric theory, the refractive index depends on the
density and on the polarisability of the atoms in a given material [72]. The
polarisability of a microporous adsorbent determines its disperse properties and,
28
thus, it is possible to calculate the Hamaker constant from knowledge of the refractive
index [41]. In [73] refractive indices of multicomponent silicate melts containing
alumina have been determined employing a high temperature elipsometer. The data
were used to propose a predictive equation for the refractive indices of silicate melts.
In [72] 13 SiO2 polymorphs with topologically different tetrahedral frameworks have
been investigated in order to find a relationship between refractive index and density.
The mean refractive indices of the guest free porosils were determined by the
immersion method using the NaD line. One example is the ZSM-5 type zeolite
silicalite-1 examined by Marler [72]. Another possibility is the method of Becke lines.
A Becke line is a bright halo near the boundary of a transparent particle, which
moves with respect to that boundary when the microscope is moved up and down.
The halo will always move up the higher refractive index medium as the position of
the focus is raised. The halo crosses the boundary to the lower refractive index
medium when the microscope is focused in downward direction. The results are
compared with pure, isotropic crystalline solids with known refractive indices (e.g.
NaCl, CaCO3). Van der Hoeven measured with this method the refractive index of
zeolite MS4A [41].
There are several approaches for quantitatively describing the refractive index. Three
of them will be explained in this study. Neglecting interactions between the atoms the
classical dielectric theory leads to the Newton-Drude relation
.1
0
2
constMNn
solid
Asolid
app
solid ==−
εα
ρ (3.32e)
where Msolid is the molar mass of the solid in kg/mol, solidα the mean average
polarisability in (C2*m2)/J, solidn the mean refractive index, 0ε the permittivity of free
space in C/(V*m) and ρapp the apparent density of the solid in kg/m3.
The second approach takes into account interactions between the atoms and the fact
that the dielectricum is not continuous. This is known as the Lorentz-Lorenz
assumption. It leads to the Clausius-Mosotti equation of nonpolar but polarisable
molecules (for a detailed derivation of this theory see Arndt and Hummel):
29
( ) solid
Asolid
app MN
nn
02
2
321
εα
ρ=
+−
(3.33)
The assumption of point charges is valid only for ideal ionic solids in which the ions
show no overlap of their electron distribution. To overcome this shortcoming Marler
suggests a general refractivity formula which takes into consideration an overlap field
of near-neighbour interactions.
( )( ) solid
Asolid
app MN
nbn
02
2
4141
επα
ρπ=
−+−
(3.34)
with the electronic overlap parameter
γπ −=3
4b (3.35)
where γ is a material constant resulting from the overlap field. Equation (3.32e)
includes as special cases the Clausius-Mosotti formula for 3
4 π=b if there is no
overlap field and the Newton-Drude equation for b = 0 if the overlap field compensates
for the Lorentz field exactly.
In this study the Clausius-Mosotti equation is used for further investigations because
by applying this approach the refractive index of structures like aluminosilicates can be
calculated a priori from knowledge on the properties of the solid. Therefore, the theory
of Clausius-Mosotti will be explained in more detail.
3.3.5.1. The Clausius-Mosotti equation
The electrical field →E can be expressed by the force which results from a small point
charge in this field. The field intensity →E is a vector in direction of the force vector
→F ,
the charge q is a scalar quantity [95]. It is
30
qFEr
r= (3.36)
For the displacement of a charge q in a homogenous field between two charged
parallel plates of distance d and voltage U the work
qUtIUdFW === (3.37)
is necessary where t is the time and I the current. With equation (3.36) if follows
dUE = (3.38)
The charges bonded on a loaded matter (e.g. the plates of a condenser) determine
the quantity of the field intensity. The charge density σ is defined as the ratio of the
matter�s charge q to its loaded surface S with
ESq
0εσ == (3.39)
where ε0 denotes the permittivity of free space, ε0 = 8.85.10-12 C/(V*m). If an electrical
nonconducting material is inserted into the electrical field the charge density rises
with a factor εr which is the relative permittivity. It is then
Er 0εεσ = (3.40)
The difference of the charge densities after and before inserting the so-called
dielectric into the electrical field is the electrical polarisation P with
( )1000 −=−= rr EEEP εεεεε (3.41)
31
Through the polarisation the charges of the dielectric are separated with the results of
a dipole moment µ. With the number density ρj of dipole moments in an examined
volume element the polarisation also can be written as [94]
µρ jP = (3.42)
It is common knowledge that the number density ρj of particles in a volume element
can be calculated via
MN appA
j
ρρ = (3.43)
where NA is the Avogadro constant, M the molar mass and ρ the density. Inserted
into equation (3.42) a relation between a molecular property, the dipole moment, and
a macroscopic measurable quantity, the polarisation P is obtained with
µρM
NP appA= (3.44)
The induced free surface charges are the consequence of three processes in the
molecular area which occur in presence of an outer electrical field. The polarisation is
then the sum of three different parts with
µPPPP ae ++= (3.45)
where Pe is the electronic polarisation, Pa the atomic polarisation and Pµ the
orientational polarisation. The electronic polarisation is evident in all atoms and
molecules. It is based on the fact that the charge centre of core frame and electronic
shell is dispersed through the outer electrical field in different directions. The atomic
polarisation occurs at all polar molecules and is based on the changes of bonding
distances and bonding angles. These two polarisation parts are combined under the
term of displacement polarisation. The induced dipole moment is the consequence of
it. The orientational polarisation occurs only at polar molecules with permanent dipole
32
moment. It is based on the partial arrangement of the molecular dipole moments in
direction of the electrical field.
In the following a molecule is considered which is not an electrical dipole in an
electrical field. Due to displacement of its charge centre it gets an induced dipole µind
which is proportional to the electrical field intensity E with
Eind αµ = (3.46)
where α is the molecular polarisability in (C2*m2)/J, a measure for the ability of the
molecule to deform. With equation (3.44) it follows that
EM
NP appA α
ρ= (3.47)
If the medium consists of polar molecules the molecules are influenced not only by
the outer electrical field Eouter but also by an inner electrical field Einner. Lorenz [128]
and Lorentz [129] independently derived the overall field E inside a spherical volume
element according to
031
εPEEEE electricalinnerouter +=+= (3.48)
With equation (3.47) it follows that
+=
031
εα
ρ PEM
NP electrical
appA (3.49)
With equation (3.41) we obtain
( )
+
−=
00 31
1 εεεα
ρ PPM
NP
r
appA (3.50)
33
This equation is independent of the polarisation and can be rewritten as the Clausius-
Mosotti equation:
( )( )2
13 0 +
−=r
r
app
A MNεε
ρεα (3.51)
With this relationship and the relative permittivity the polarisability of the material can
be calculated or vice versa. The aim of the next Section is to derive a new
relationship in order to calculate the refractive index of solids like zeolites. It is based
on the theory of Maxwell.
3.3.5.2. Refractive index of zeolites
Electromagnetic waves are generated when charged particles accelerate. Their
existence was predicted in 1865 by James Clerk Maxwell on theoretical grounds.
They were first produced and studied in 1888 by Heinrich Rudolf Hertz. The Maxwell
equations are a group of equations which centralise the basic laws of electric and
magnetic appearances to one package. Only the last of four equations has been
derived by Maxwell but he was the first to draw the right conclusions from them [96].
He found that the electromagnetic waves spread out with a velocity:
000
1µε
=c (3.52)
With equation (3.52) Maxwell has unified the permittivity of free space ε0, the
permeability of free space µ0, and the speed of light in vacuum, which was found to
be c0.
In the general case the spreading speed of the electromagnetic wave in a medium is:
µε1=w (3.53)
34
where ε is the permittivity of the medium and µ the permeability of it. It is
0εεε r= and (3.54)
0µµµ r= (3.55)
where εr and µr are the relative permittivity and the relative permeability of the
medium respectively. With equation (3.53) it follows:
00
1µµεε rr
w = (3.56)
Especially for visible light µr can only approximately be 1 because matters with a big
µr, e.g. ferromagnetics like iron and nickel, are intransparent. Zeolites are assumed in
this work not to be magnetic. With equation (3.52) and the above assumption it
follows:
r
cwε1
0= (3.57)
The relation of this equation with the refractive index n can be found by using the law
of Snel. Willebrord Snel van Royen, a Dutch mathematician and geodesist,
discovered in the early 1600s the law of refraction. If an electromagnetic wave
trespasses the frontier of two media it changes with its velocity also its spreading
direction. The wave will be refracted like in figure 3.7.
The law of refraction predicats that the ratio of the angles α and β are equal to the
ratio of the wave velocities in the different media, therefore
( )( ) n
ww ==
2
1
sinsin
βα (3.58)
35
Figure 3.7: Snel�s law of refraction.
This ratio is defined as the refractive index n. With equation (3.57) and w1 = c0 in this
special case and with w2 as the wave velocity in the medium, it is
rwc
wwn ε=== 0
2
1 (3.59)
known as the relation of Maxwell. If the Maxwell equation
2nr =ε (3.60)
is inserted in the Clausius-Mosotti equation (3.51) the Lorenz-Lorentz relationship
between the refractive index n of a medium and the polarisability α is obtained:
( )( )2
13 2
2
0 +−=
nnMN
app
A
ρεα (3.61)
The aim of this Section is to derive a formulation to calculate the refractive index of
solids like zeolites. Therefore, the polarisability of the adsorbent or the left side of
equation (3.61) has to be known. This left side is also defined as the molar refractivity
Rm with
( )( )2
13 2
2
0 +−==
nnMNR
app
Am ρε
α (3.62)
where Rm is in m3/mol [98]. The refractive index of the solid is then
Medium 1 with velocity w1
Medium 2 with velocity w2
α
β
inlet wave
refracted wave
perpendicular
36
MRM
R
nappm
appm
ρ
ρ
−
+=
1
21 (3.63)
The molar refraction of a molecule with the structure BA xx BA is
BmBAmAm RxRxR ,, += (3.64)
where xA and xB are the stoichiometrical coefficients and Rm,A and Rm,B the ionic
molar refractions of the matters A and B, respectively. In general it is
∑=i
imim RxR , (3.65)
The ionic molar refractivities Rm,i can be calculated via equation (3.62) if the ionic
polarisabilities αi are known. It is therefore
iA
imNR αε 0
, 3= (3.66)
In the Handbook of Chemistry and Physics [6] the polarisability volumes 'iα are
tabulated for the interesting ions of aluminosilicates. The polarisability αi is related to
the polarisability volume 'iα via
'
04 ii αεπα = (3.67)
In table 3.6 the interesting properties are listed. As an example the refractive index of
MS5A is calculated.
37
Table 3.6: Numerical values for the ionic molar refractivities.
Ion Polarisability volume 'iα in 10-30 m3 [6]
Polarisability
αI in 10-40 (C2*m2)/J
with equation (4.34)
Ionic molar refraction in
10-6 m3/mol calculated
with equation (4.33)
Na+ 0.179 0.1991 0.4515
Ca2+ 0.470 0.5229 1.1857
Si4+ 0.0165 0.0183 0.0416
Al3+ 0.0520 0.0578 0.1311
K+ 0.830 0.9235 2.0938
O2- 3.880 4.3171 9.7881
MS5A has the chemical structure Ca5Na2((AlO2)12(SiO2)12). Its molar mass M is
calculated due to the molar masses of the involved matters which are listed in table
3.7 with
∑=i
iiMxM (3.68)
where M and Mi is in kg/mol and xi the stoichiometrical coefficients. For MS5A it is
molkgM AMS 68.15 = (3.69)
Sievers [37] measured for MS5A a pellet density of
35, 1150mkg
AMSapp =ρ (3.70)
38
Table 3.7: Molar masses of interesting matters.
Compound Molar mass in 10-3
kg/mol
Na 22,989
Ca 40,08
Si 28,086
Al 26,981
K 39,102
O 15,999
The molar refractivity of MS5A is calculated with the values of the ionic molar
refractivites of table 3.6 with
molmR AMSm
34
5, 1079.4 −⋅= (3.71)
Finally the refractive index of MS5A is
57.15 =AMSn (3.72)
Direct measurements for zeolites were only found for MS4A and silicalite-1.
Nevertheless the proposed way of refractive index calculation is very straightforward
and independent of experimental work. It is an important parameter for the
calculation of Hamaker constants, e.g. for aluminosilicates, as shown in the next
Section.
39
3.3.6. Hamaker constant
For the characterisation of solid materials their interaction potentials with other solids or
fluids are of strong interest. One of the most important parameters for describing the
disperse energies of a given system is the so-called Hamaker constant of the solid.
This constant is also eminent for the calculation of adsorption equilibria of gases on
energetically homogeneous and heterogeneous surfaces.
Hamaker constants of organics (solids and liquids) in air have magnitudes of
(6 ± 1).10-20 J and are reported to be lower than those of inorganic salts ((9 ± 3).10-20
J), which are again lower than those of metal oxides ((13 ± 4).10-20 J) [41]. Götzinger
[49] reports in table 3.8 constants of several inorganic and organic matters.
Table 3.8: Hamaker constants of inorganic and organic matters.
Matter Hamaker constant in 10-20 J
Al2O3 12.19 � 17.01
TiO2 16.05 � 22.08
Silicium 20.60 � 22.90
SiO2 6.28 � 6.90
Polystyrol 6.34 � 8.0
Graphite 12.71 � 28.0
Active carbon 6.0
Van der Hoeven [41] reports two methods for the estimation of Hamaker constants.
The first one is based on the macroscopic theory of Lifshitz. It requires the knowledge
of the absorption spectrum and the dispersion, i.e. the refractive index n(ω) and the
dielectric permittivity ε(ω) as a function of frequency. The other method is the
Hamaker-de Boer microscopic summation approach. This method is based on the
subdivision of the inorganic solid molecules into atoms, which all have an average
composition and polarisability, identical to that of the solid molecule. The method
requires, apart from the refractive indexes and dielectric constants, the molar masses
40
and the specific densities of the materials involved. A brief description of these two
methods and their theories is given in the next sections.
3.3.6.1. Macroscopic approach of Lifshitz
Hamaker [75] experienced in 1937 the existence of adhesive forces between small
particles. He ascribed this adhesion for a large part to London van-der-Waals forces. In
his theory the energy of interaction between two particles containing ρj,at atoms per m3
is given by
∫ ∫−=1 2
6
2,
21V V
atj
rdvdvE
βρ (3.73)
where dv1, dv2, V1 and V2 designate volume elements and total volumes of the two
particles, respectively, r denotes the distance between dv1 and dv2 in m and β is the
London van-der-Waals constant in J*m6.
The mutual energy of two spheres of diameters D1 and D2, with the distance d
according to figure 3.8 and containing ρ1,at and ρ2,at atoms per m3 which interact with
an energy 6rβ is according to Hamaker (see Figure 3.8)
+++++
++++
+++
−=1111
21
1112
1
11112
1
1
1112
1
1 ln2121
yxyxxxyxx
yxyxxy
xyxxy
HaE (3.74)
where the variables are defined as
βρρπ atatHa ,2,12= , (3.75)
11 D
dx = and (3.76)
41
1
21 DDy = (3.77)
Figure 3.8: Two interacting spheres.
The constant Ha is also known as the Hamaker constant and is defined for like
particles with the same atom density atj ,ρ as
βρπ 2,
2atjHa = (3.78)
When x1 << 1 or d → 0 equation (3.74) is approximated by
( )12
12
121
DDdDDHaE+
−= (3.79)
Ackler et al. [76] calculated the Hamaker constant of muscovite mica, Al2O3, SiO2,
Si3N4 and rutile TiO2 from Lifshitz theory. He compared these constants to values
calculated from physical properties using the Tabor-Winterton approximation.
Lifshitz developed a theory for the nonretarded case where the interparticle
separations are small enough that the interactions between dipoles is considered
instantaneous. The van-der-Waals interaction is the result of fluctuations in the
electromagnetic field between two macroscopic bodies, modified by the separating
media, where the interaction can be referred to the standing waves which only occur
at certain frequencies. Hence, the van-der-Waals interaction and, thus, the
associated Hamaker constants can be estimated from the knowledge of the
frequency dependent dielectric properties of the interacting materials together with
D1 D2
d
42
the intervening medium and the geometry of the bodies. The accuracy of the
estimated Hamaker constants is directly related to the precision and accuracy of the
dielectric spectra and the mathematical representation of this data [77].
The dielectric properties of materials are commonly represented by the frequency
dependent dielectric response function
( ) ( ) ( )ωεωεωε ′′+′= i (3.80)
which is a complex function here, ( )ωε ′ is the real part and ( )ωε ′′ the imaginary part.
For a static applied field (ω = 0), non-conductor materials are characterised
by ( )0ε ′′ =0, hence ( )0ε ′ = ( )0ε which is the static dielectric constant. For this case the
real part of the dielectric response is directly related to the refractive index n through
( ) ( ) ( )000 2n=′= εε (3.81)
Dielectric data can be obtained with a number of methods including capacitance
bridge, optical reflectance and electron loss spectrometry. Most of these techniques
yield only one of the components of the complex dielectric response function which
often necessitate the transformation from the real to the imaginary component or vice
versa. Such a transformation can be performed using the Kramers-Kronig relations,
one of them defined as
( ) ( )∫∞
−′′
+=′0
1221
121 dxxx
ωωε
πωε (3.82)
However, since the Kramers-Kronig relation is formally correct only when ( )ωε ′′ is
known in the entire frequency range, 0 < ω < ∞ , this procedure is applicable. Nimhan
and Parsegian showed that by constructing an imaginary dielectric response function,
ε(iξm), a much simpler dielectric representation could be used for the purpose of
calculating Hamaker constants. In the study of Bergström [77] ε(iξm) is represented for
most of the inorganic materials by one UV and IR relaxation through
43
( ) 22m
11
1i
+
+
+
+=
UV
UV
IR
IR CC
ωξ
ωξ
ξε (3.83)
where each material is characterised by four parameters; CIR and CUV are the
absorption strengths in the IR and UV range, respectively, and ωIR and ωUV represent
the characteristic absorption frequencies in the IR and UV range, respectively. It is
j
jj
fC
ωπ2= , j = IR, UV (3.84)
where fj is the oscillator strength and ωj is the relaxation frequency of the absorption
band. Now, the non-retarded Hamaker constant of two particles 1 and 2 in medium 3
can be derived from the integral
( )∑∫∞
=
∞
−∆∆−=0 0
111132 )exp(23131ln2
3m
dxxxkTHa (3.85)
This integral can be solved analytically to yield
( )∑∑∞
=
∞
=
∆∆=0 1
31322313
23
m s
s
skTHa (3.86)
where k is the Boltzmann constant and T the temperature in K and with the
differences in dielectric response, kl∆ , defined as
( ) ( )( ) ( )mlmk
mlmk
iiii
klξεξεξεξε
+−
=∆ (3.87)
Bergström [77] used simplified approximations by only retaining a single UV
relaxation to represent the dielectric response of each material. His simplifications
have been motivated by the important role of the ultraviolet relaxations on the
magnitude of the disperse interactions. With such a simple dielectric representation
44
each material is fully described by three parameters, ε(0), the characteristic
frequency in the ultraviolet, ωUV, and the low-frequency limit of refractive index in the
visible UV range, n0 (since 12 −= visUV nC ). For example he calculated the Hamaker
constant Ha131 for two identical materials 1, interacting across a medium 3, by using
(3.36), neglecting all summation terms s > 1, converting the summation over m to an
integral for n > 1, and finally setting ω1 = ω3 = ω. This results in
( ) ( )( ) ( )
( )( ) 2
323
21
223
21
2
31
31131 232
30000
43
nn
nnhkTHa+
−+
+−
=π
ωεεεε (3.88)
Both authors (Bergström and Ackler) compared their data with the Tabor-Winterton
approximation for the Hamaker constant. Beside the above simplification of ignoring
contributions from vibrations in the infrared a second approximation is made that the
absorption in the UV occurs within a narrow frequency range. It follows
( )( ) 2
323
21
223
21
131 283
nn
nnhHa e
+
−=
νπ (3.89)
where h is the Planck�s constant and eν the main electronic absorption frequency of
about 3*1015 Hz also called the plasma frequency. Recalling that n = ε2, it is apparent
that the Tabor-Winterton approximation enables the calculation of a Hamaker constant
only from physical properties such as the refractive index or the dielectric constant of
the materials of interest and of the medium �3�.
In the case of two interacting identiacl macrobodies (index1) in vacuum Israelachvili
derived the following expression which allows to approximate the Hamaker constant
[21, 98]:
( )( )
( )( ) 2
321
221
2
1
111
1
1216
32010
43
+
−+
+−
=n
nhkTHa eνεε (3.90)
45
In this equation, instead of integration over the whole frequency area, ε(ω) and n(ω)
in the original equations of Lifshitz are replaced by two static values, one at low
frequencies, i.e. the static dielectric constant ε, and one at high frequency, i.e. the
refractive index n of the medium in visible light.
In this macroscopic approach the static dielectric constant is not known in many
cases or must be measured in a complex way. Considering the large differences in
chemical compositions, it is unlikely that inorganic matters all have the same νe value
of 3.1015 s-1. Nevertheless Israelachvili has shown that the results of equation (3.90)
are accurate within a few percent when compared with the results of exact
computations.
In the next Section the Hamaker � de Boer microscopic approach is introduced which
is used in this work for the derivation of the London interaction forces.
3.3.6.2. Microscopic approach of Hamaker � de Boer
The Hamaker � de Boer microscopic approach is based on the equation
βρπ 2,
2atjHa = (3.78)
where ρj,at is the is the number density of solid atoms in 1/m3 and β the London
constant in J*m6 [75]. This approach is strictly speaking valid only for monoatomic
solids. If the Hamaker constant of multiatomic solids is calculated their chemical
structure has to be considered. The proposed route is based on a subdivision of the
solid molecule into atoms, which all have a mean concentration and polarisability.
The Hamaker constant is then calculated due to these mean polarisabilities. A
second assumption is that only the valence electrons of these atoms contribute to the
disperse interaction potential. So the average atoms are all bonded by semi-polar
bonds and only the bond-forming electrons contribute to the higher energetic waves.
The Hamaker constant for multiatomic solid molecules is then
46
βρπ 2atj,
2=Ha (3.91)
where ρj,at is the number density of solid atoms and β the London constant. The
number density can be calculated via
MN
n appAaatj
ρρ =, (3.92)
where na is the number of atoms in the molecule. M denotes the relative atomic mass
related to the same solid molecule with which the number of atoms is calculated. In
the following the zeolite MS5A with the chemical structure Ca5Na2((AlO2)12(SiO2)12) is
examined. Its molecular mass is given in equation (3.69). The overall number of
atoms na in this molecule is
795, =AMSan (3.93)
With the density in equation (3.70)
328
,511027.3matAMS ⋅=ρ (3.94)
is obtained.
The magnitude of this number density of atoms for zeolite MS5A is comparable to
many materials, for example silicon has a number density of 328 1100.5m
⋅ and SiO2 a
number density of 328 1102.2m
⋅ [101].
In 1931 Slater and Kirkwood [100] derived an equation for the calculation of the
London constant. Van-der-Hoeven [41] modified this equation by introducing an
atomic polarisability. The London constant for macro bodies like zeolites can be
calculated by
47
2
00 44
3
=
επανβ hs (3.95)
The letter s represents the number of electron bondings per atom. It is obtained
simply from the summation of the single and double bond forming electrons involved
in atomic bonding, divided by the number of atoms in the molecule. The parameter s
is only reasonable for the calculation of the Hamaker constant and London dispersion
potential, respectively. Later this potential will be introduced as the induced dipole �
induced dipole interaction energy.
For example, for zeolite MS5A (Ca5Na2((AlO2)12(SiO2)12)) the total number of atoms
in the molecule is 79. The number of valence electrons is for calcium 5*2 = 10, for
sodium 2*1 = 2, for aluminium 12*3 = 36, for silicon 12*4 = 48 and for oxygen 48*6 =
288, altogether 384 valence electrons. The number of electron bondings is then 192.
It follows for s that
43.279
1925 ==AMSs (3.96)
The polarisability of the solid atoms can be calculated from the refractive index using
the Lorenz-Lorentz equation
( )( )
( )( )2
1321
34
4 2
2
02
2
0 +−=
+−=
nn
NnM
nn
Nn
M
AaAa
ρε
ρπεπα (3.97)
with the already known parameters.
The variable ν0 is the frequency of the electron in the ground state in 1/s. It is
απν 1
20em
e= (3.98)
48
where e is the charge of an electron (e = 1.60*10-19 C) and me the mass of an
electron (me = 9.11*10-31 kg). With known or calculated values of the refractive index
n, the apparent density ρ, the molar mass M and the chemical structure of the solid
the Hamaker constant for zeolites can be calculated. For MS5A it is:
JHa AMS20
5 1066.7 −⋅= (3.99)
3.4. Adsorption isotherms
In this study single-component equilibria of gases or vapours with microporous solids
are examined. The context does not cover phenomena like chemisorption, hydrogen
bonding, steric or formselective effects. Attention is payed only to physisorption only,
the adsorptive gas molecules are physically bonded to the microporous surface via
potential forces. In this Section the existing types and common descriptions of
adsorption isotherms are discussed which are used in this work for different
purposes. E.g. measuring the specific surface according to the BET isotherm,
calculating Henry coefficients by means of Langmuir and Toth isotherms.
3.4.1. Types of isotherms
The first classification of physical adsorption isotherms was presented by Brunauer et
al. [78]. In 1985 the IUPAC Commission on Colloid and Surface Chemistry proposed
modifications of this classification by adding a sixth type, the stepped isotherm, to the
original five types of Brunauer et al.. Type I (the Langmuir isotherm) is typical for
microporous adsorbents (activated carbons, zeolites). The next two are typical for
nonporous materials with strong (type II) and weak (type III) fluid-surface forces. Types
IV and V are characteristic for mesoporous materials when capillary condensation
occurs (these types exhibit hysteresis loop). Type VI occurs for materials with relatively
strong fluid-surface forces, usually when the temperature is near the melting point of
the adsorbed gas [79]. Figure 3.9 shows the principle types of adsorption isotherms
regarding to IUPAC classification.
49
Figure 3.9: Principle types of adsorption isotherms regarding to IUPAC classification.
Nevertheless, the classification of adsorption isotherms in several types is not helpful
for the understanding of the decisive physical processes. Also the evaluation of the
isotherm types as advantageous or disadvantageous is ambivalent, because an
advantageous form for adsorption is at the same time a disadvantageous form for
desorption [24]. This work deals mainly with isotherms of type I.
n(mol/kg)
p(Pa)
n(mol/kg)
p(Pa)
type I type II
n(mol/kg)
p(Pa)
type III
n(mol/kg)
p(Pa)
type IV
n(mol/kg)
p(Pa)
type V
n(mol/kg)
p(Pa)
type VI
50
3.4.2. Langmuir isotherm
Irving Langmuir developed 1918 a quantitative theory of monolayer adsorption which
is simple and ingenious at the same time [80]. Figure 3.10 shows the principle basis
of the Langmuir model. The assumptions are [82]:
- The molecules are adsorbed on well-defined and localised places.
- Each adsorption place can bind only one adsorptive molecule.
- All adsorption places are energetically equivalent.
- There are no interactions between adsorbed molecules.
Figure 3.10: Principle of the Langmuir model.
For the attachment of molecules with solid surfaces the following reaction is
essential:
OPGads
des
k
kf ↔+∆ (3.100)
The adjustment of a sorption equilibrium leads to the equilibrium constant KL, which
is the ratio of the concentration of free gas molecules [G], the concentration of free
places [∆f] and the concentration of occupied places [OP]:
occupied place OP = (1-∆f) free place ∆f
gas molecule G solid
surface S
pressure p of gas continuum
51
[ ][ ] [ ]fL GOPK
∆= (3.101)
KL is called the Langmuir constant and has the dimension 1/pressure because the
concentration of free gas molecules [G] is equivalent to the partial pressure p of the
adsorptive. The sum of the concentration of free places and occupied places is the
maximum available concentration of sorption places [SP]. It is
[ ] [ ] [ ]fOPSP ∆+= (3.102)
and it follows with
[ ] [ ] [ ]OPSPf −=∆ (3.103)
for the Langmuir constant in equation (3.101)
[ ][ ] [ ] [ ]( )OPSPG
OPKL −= (3.104)
After resolution to the concentration of occupied places it follows:
[ ] [ ] [ ][ ]GKGKSPOPL
L
+=
1 (3.105)
The concentrations of occupied and maximum available sorption places are
equivalent to the loading n and the maximum available loading nmon of gas molecules
on the solid surface. So the Langmuir isotherm for monolayer loadings can be
immediately written to
pKpKnnL
Lmon +
=1
(3.106)
52
For very low pressures KL*p becomes negligible compared to unity, so that as a first
approximation the Henry isotherm is obtained:
pKnn Lmon= (3.107)
The product of the constants Lmon Kn * is equal to the Henry coefficient He. These
constants are adjustable coefficients to adsorption data. With a linearised form the
Langmuir coefficients can be derived via a graph with p as the x axis and p/n as the y
axis. It is
pnKnn
p
monLmon
11 += (3.108)
The Henry coefficient is then indirect proportional to the intercept point Li:
( )Lmoni KnL
He1
11 == (3.109)
The Langmuir isotherm can also be derived from statistical thermodynamics and by
using Gibb�s isotherm [81].
3.4.3. BET isotherm
The BET isotherm is eminent for the determination of the specific surface of
adsorbents. Therefore, the isotherm is derived theoretically in this Section. With the
help of a few simplifying assumptions it is possible to carry out an isotherm derivation
for multimolecular layers that is similar to Langmuir�s derivation for monolayers [62].
Brunauer, Emmett and Teller look at the surface layers si covered by only i layers of
adsorbed molecules. Using Langmuir�s equation for monolayer adsorption they
derive the general equation
ℜ−
=−ads
iiiii T
Hsbpsa exp1 (3.110)
53
where p is the pressure in Pa, Hi the specific heat of the i-th layer in J/mol, ai a mole
rate constant of layer i in mol/(m2*s*Pa), bi a mole rate constant of layer i in
mol/(m2*s) and si the surface of the molecule layer i in m2. Equation (3.110) means
that the rate of condensation on top of the layer i-1 is equal to the rate of evaporation
of the i-th layer. Brunauer et al. proposed the following points:
- The specific heat of the first layer is the heat of adsorption Hads.
- The specific heat of the i-th layer is the heat of liquefaction HL with
Li HHHH === ...21 (3.101)
This is equivalent to saying that the evaporation-condensation properties of
the molecules in the second and higher adsorbed layers are the same as
those of the liquid state. This means also that
gab
ab
ab
i
i === ...3
3
2
2 (3.102)
where g is an appropriate constant in Pa.
- At equilibrium the surface layers si remain constant.
The total surface S is given by
∑∞
==
0iisS (3.103)
and the total volume adsorbed is
∑∞
==
00i
iG sivV (3.104)
where v0 is the volume of gas adsorbed on one square meter of the adsorbent
surface when it is covered with a complete monolayer of adsorbed gas. It follows that
54
∑
∑∞
=
∞
===
0
0
,0
ii
ii
monG
GG
s
si
VV
vSV (3.105)
where VG,mon is the volume of gas adsorbed when the entire adsorbent surface is
covered with a complete monolayer. One can express now s1, s2, � si in terms of s0
whereas
ℜ
==ads
ads
TH
pba
ywhereyss exp1
101 (3.106)
ℜ
==ads
L
TH
gpwheress exp12 ϕϕ (3.107)
ℜ
−====== −−
−ads
Ladsiiiii T
HHbgaycwherescsysss exp1
100
11
11 ϕ
ϕϕϕϕ (3.108)
The variable ϕ can also be understood as the relative saturation, the ratio of the
pressure of the gas to its saturation pressure. The variable c is the well known BET
constant.
Substituting (3.106), (3.107) and (3.108) into equation (3.105) yields
+=
∑
∑∞
=
∞
=
1
10
, 1i
io
i
i
monG
G
cs
ics
VV
ϕ
ϕ (3.109)
If the number of the adsorbed layers cannot exceed a maximum number n, then the
summation of the two series in equation (3.109) is to be carried out to n terms only,
and not to infinity. It follows then
( )( )( ) 1
1
, 1111
1 +
+
−−+++−
−= n
nn
monG
G
ccnnc
VV
ϕϕϕϕ
ϕϕ (3.110)
55
which is the well known general BET isotherm. For the determination of the solid
surface SBET the summation of the two series has to be carried out to infinity. The
denominator sum can be represented as the sum of an infinite geometric progression
ϕϕϕ−
=∑∞
= 11i
i (3.111)
and the numerator sum can be expressed also in the same way as
( )21 1 ϕ
ϕϕ−
=∑∞
=i
ii (3.112)
It follows therefore that
( )( )ϕϕϕϕ
cc
VV
monG
G
+−−=
11,
(3.113)
With
0pp=ϕ (3.114)
equation (3.2) for the determination of the solid surface SBET is obtained. 3.4.4. Toth isotherm
One of the assumptions of Langmuir is the energetic homogeneity of the adsorbent
surface. This assumption is not always valid. Toth also tried to take into account the
energetic heterogeneity in his adsorption equilibrium modell [83]. The basis of his
theory is the dividing of the surface in a huge number of homogeneous fields with
very small areas. With the partial loading
56
monnnN = (3.115)
he wrote in a first step the Langmuir equation (3.106) in the differential form
pKdNdp
pN
L=−1 (3.116)
Toth introduced a differential function Ψ(p) which describes the whole energetic
heterogeneous loading interval:
( ) mT pKp =ψ (3.117)
where 0 < m < 1 and KT is the Toth constant in (1/Pa)m. The backwards integration of
equation (3.116) with equation (3.117) yields the Toth isotherm
mm
T
pK
pnnN 1max 1
+
== (3.118)
The Henry coefficient follows at very small pressures. It is
( )mTpKn
pnHe
1
max0lim =
=
→ (3.119)
The qualitative difference of Henry coefficient derivations via the Langmuir and Toth
isotherms is discussed in Chapter 7.
3.4.5. Potential theories
The isotherms of Langmuir, Brunauer et al. and Toth are derived from equilibrium or
kinetic considerations. Another possibility is the inclusion of adsorption forces or
potentials in the models. In this Section classic and modern concepts of potential
theories are presented.
57
3.4.5.1. Polanyi and Dubinin
Polanyi developed his potential theory in 1916 [84]. The assumptions made are:
- The solid concentrates the adsorbed material on its surface by means of
attraction forces.
- The attraction forces are dependent on system properties and the distance
of the interacting molecules.
- The attraction forces are independent of temperature.
- There is no dissolution of the adsorbed molecules in the adsorbent.
- The adsorption volume is filled by liquid adsorbate only.
- Gas phase exhibits ideal gas behaviour.
- The liquid adsorbed phase is incompressible.
- Creating a liquid surface involves negligible work.
The adsorption potential is defined by Polanyi as the isothermal compression work of a
molecule with partial pressure p which transfers from the gas phase to a point in the
adsorbed phase with saturation pressure p0 with
ℜ=
ppTads
0
lnε (3.120)
The later studies of Dubinin on adsorption equilibria are based on the potential theory
of Polanyi. The adsorption in microporous adsorbents is understood as successive
filling of the pore volume. The principle task in his theory consists of finding the
distribution of the volume adsorption space, v, as a function of Polanyi adsorption
potential ε with
=
βεfv (3.121)
58
where β is called the affinity coefficient. The forces of attraction of the molecules to
the surface of the adsorbent are related to the ratio of polarisability of the vapourous
molecules, therefore
refref αα
εεβ == (3.122)
where α is the polarisability of the molecule and the index �ref� indicates the
reference molecule. Dubinin assumes that the temperature does not produce any
effect on this characteristic curve v, each adsorbent has its own characteristic curve
that describes its adsorption-force field. For adsorbents of type Ι, including active
carbons and zeolites with extremely small micropores, the equation of the
characteristic curve assumes the form
−= 2
2
max expβεκvv (3.123)
where vmax is the limiting volume of adsorption space that equals the volume of all
micropores. The parameter κ reflects the function of distribution of volume of the
pores according to sizes. This theory was tested on the basis of numerous
substances. It is very successful in describing adsorption equilibria of active carbons.
The advantage of this approach is that for the description of the characteristic curve
only one isotherm has to be examined experimentally over a big concentration
interval [82]. With equation (3.120) in equation (3.123) the isotherm of Dubinin-
Raduskevich is given by
max
20
max
lnexpnn
ppT
vv ads =
ℜ−=
βκ (3.124)
where the ratio of the volumes can be replaced by the ratio of the loading n/nmax for a
constant density of the adsorbed phase.
A general form of equation (3.124) was derived by Dubinin and Astakhov [130] with
59
max
0
max
lnexpnn
ppT
vv
Yads =
ℜ−= ∗
βκ 1 < Y < 3 (3.125)
It is not possible to transfer equation (3.124) and (3.125) to Henry�s-law in the low
partial pressure interval. The postulation for the thermodynamic consistency of a
isotherm description is not fulfilled. Nevertheless, for bigger concentrations the
Dubinin isotherms are powerful instruments.
3.4.5.2. Monte Carlo simulations
The objective of molecular modelling systems is to predict macroscopic,
experimentally observable properties by simulating the molecular motion of materials
[87]. One starts with a model that describes the inter- and intra-molecular interactions
and then relates this through statistical mechanics to macroscopic properties. These
can range from thermodynamic properties to transport properties and to molecular
arrangements.
There are two major ways to solve these problems statistically. In the first one,
molecular dynamics (MD), molecules move under the influence of their own
molecular forces. In micro canonical molecular dynamics, one specifies the number
of molecules, the energy in the system and the volume of the system. The positions
of each molecule are then calculated by solving Newton�s equation for the force F,
the mass m and the acceleration a:
amF = (3.126)
From this, the desired properties can be gained as a function of position and velocity
as a function of time.
Another way is the Monte Carlo simulation which is a stochastic method. In a
traditional or canonical Monte Carlo simulation, the volume, the number of molecules,
and the temperature of the system are specified. The Monte Carlo method uses a
60
random number generator to propagate molecules according to the canonical
probability distribution. Translation and rotation motions are accepted or rejected
according to canonical accepted rules. By examining the accepted and rejected
motions, an average of them can be made in order to obtain equilibrium properties for
the system of interest. The algorithm in figure 3.11 gives the steps involved in Monte
Carlo simulations.
Figure 3.11: Steps involved in Monte Carlo simulations.
In the grand canonical Monte Carlo (GCMC) ensemble the number of particles
fluctuates while the chemical potential is constrained. This makes GCMC a natural
choice for studying phase equilibrium since the chemical potential of two phases
must be equal at equilibrium. Thus, GCMC uses the chemical potential as the basis
for adding and removing molecules from the system. In addition, the solid phase is in
contact with a vapour phase of known fugacity, which is related to the chemical
potential. In GCMC the four basic motions of a particle are as follows:
Specify number of molecules, volume and temperature
Create the initial configuration
Compute the energy of the system
Perturb the system
Compute the new energy of the system
Accept or reject based on acceptance rules
61
- Insertion into a random position in the system.
- Deletion
- Rotation
- Translation into a new position in the system.
In their review Smit and Krishna discuss the use of modern Monte Carlo simulations
in the context for simulating the properties of molecules adsorbed in zeolites [88].
They show that it is possible to use Monte Carlo simulations to obtain a reasonable
estimate of an adsorption isotherm of pure hydrocarbons and their mixtures in
zeolites. As in all kinds of simulation methods the restrictions of Monte Carlo
simulations are the limits in calculation capacity. Also the results have to be verified
with experimental data because it is only a random model and not an a priori one.
Nevertheless, intensive engagement with the theme of high-performance-computing
is necessary to further push this auspicious methods.
3.4.6. Isotherms based on the Hamaker energy of the solid
Maurer [26] has demonstrated that the Hamaker energy Haj of the adsorbent and the
critical data pc and Tc of the adsorptive are basic properties in order to describe
equilibrium isotherms of all gases on active carbon. Mersmann et al [27] introduced a
diagram in which the initial or Henry loading is plotted versus an interaction
parameter Cj
j
ads
C
pHa
TT
3
2πσ
, as can be seen in figure 3.12.
With respect to the entire isotherm in figure 3.12 the dimensionless numbers npHe ,
23
3
Cj
j
ads
C
pHa
TT
σand
BET
iA
SNn 2σ have been introduced. However, all these results are
restricted either to activated carbon and unpolar zeolites like silicalite-1 or to nonpolar
adsorptive molecules in the case of polar zeolites.
62
Figure 3.12: The initial or Henry loading plotted versus the interaction parameter
[27].
Dimensionless Henry Curve
1E+00
1E+01
1E+02
1E+03
1E+04
1E+05
1E+06
1E+07
1E+08
0 10 20 30
Interaction Parameter IP=
Initi
al (o
r Hen
ry) L
oadi
ng IL
IL=
(He
ℜ Τ)/(
σi,j
SB
ET)
Cj
j
ads
C
pHa
TT
3
2πσ
63
4. Experimental Methods for Adsorption Measurements
There are classical and also new techniques for measuring the adsorption equilibrium
of gases on microporous adsorbents. The classical procedures described in this
Chapter are the volumetric [8, 37, 57] and the packing bed methods [36]. The
gravimetric method is explained in Chapter 7 regarding materials and methods used
in this study. Newer techniques are the zero length column [89, 90] and the
concentration pulse chromatography methods [91, 92].
4.1. Volumetric method
This method is widely used in literature. The adsorption equilibrium time is shortened
when a gas pump is used for forced circulation of the adsorptive. Sievers [37] utilized
a circulation pump with an electro-magnetic clutch between motor and rotary slide
valve to ensure the required sealing. Figure 4.1 shows a feasible scheme of the
volumetric method.
First, carrier gas (H2) is dosed to a defined mixing volume Vmix, next a small amount
of adsorptive is fed. The pressure pbef, the temperature Tbef and the molar fraction of
the adsorptive ybef are measured, ybef with a gas chromatograph. With this data the
moles of adsorptive in the mixing volume can be measured and calculated:
adsbef
mixbefbefbef TZ
Vpyn
ℜ= (4.1)
The compressibility factor Zbef can be calculated via the method of Redlich-Kwong-
Soave which needs only critical data of pure substances. For adsorption the volume
and also the adsorbent is switched on. The overall volume is Vads. The pressure paft,
the temperature Taft and the molar fraction yaft are measured. It follows:
64
Figure 4.1: Scheme of the volumetric method.
adsaft
adsaftaftaft TZ
Vpyn
ℜ= (4.2)
The adsorbed molar quantity can be calculated as the difference of moles in the
vapour before and after adsorption:
adsaft
adsaftaft
adsbef
mixbefbefaftbefads TZ
VpyTZVpy
nnnℜ
−ℜ
=−= (4.3)
PI TI
TIC Thermostat
Isotherm adsorber
Mixing vessel
TI
offgas
vacuum pump
circulation pump Gas
chromato-graph
PI PI
offgas
H2 adsorptives 1, 2 and 3
1 2 3
He
65
The loading n of the adsorbent is the ratio of the adsorbed moles to the mass of the
adsorbent:
ads
ads
mn
n = (4.4)
The equilibrium pressure pequilibrium for the adsorption isotherm is the partial pressure
of the adsorptive in the carrier gas which is assumed not to be adsorbed. It is
aftaftmequilibriu pyp = (4.5)
4.2. Packed bed method
In the packed bed method the calculation of the equilibrium is managed by examining
the breakthrough curve of the adsorber. A mass balance is drawn around the packed
bed. The adsorbed molar amount N of an adsorptive related to the adsorbent mass in
the packed bed mbed after a time t can be calculated by
bed
outin
mNN
n−
= (4.6)
With
in
ininin M
VN
ρ= and (4.7a)
out
outoutout M
VN
ρ=
it follows
bed
outoutinin
mMVV
nρρ −
= (4.8)
66
where ρ is the partial density of the adsorptive in kg/ m3, V the gas volume in m3 and
M the molecular mass of the adsorptive in kg/mol. The law of ideal gases delivers the
following relationship
dsads
dsds
pTpT
tan
tantanρρ = (4.9)
The index stand means standard state (Tstand = 273.15 K, pstand = 1.01325*105 Pa)
and ρstand is the standard density of the adsorptive.
The gas volumes which have passed through the packed bed from t = 0 to
breakthrough can be determined via integration over the time. This calculation is
demonstrated exemplarily for the outlet stream. The outlet volume stream ( )tVout& is
composed of the carrier gas stream ( )tVC& and the adsorptive flow ( )tV adsout ,& with the
molar fraction yout,ads, so
( ) ( ) ( ) ( )tyV
VtyVtVVtVadsout
CoutadsoutCadsoutCout
,,, 1−
=+=+=&
&&&&& (4.10)
The adsorptive flow is then
( ) ( )( )( )tytyV
Vty
VVVtV
adsout
adsoutCC
adsout
CCoutadsout
,
,
,, 11 −
=−−
=−=&
&&
&&& (4.11)
An analogous equation for the inlet adsorptive flow can be written where the inlet
molar fraction of the adsorptive yin,ads is constant:
( )adsin
adsinCadsin y
yVtV
,
,, 1−
=&
& (4.12)
From equation (4.10) it follows for the carrier gas stream
67
( )adsinGC ydwV ,
2
14
−= π& (4.13)
with wG as the gas velocity in the inlet which is flow controlled and d as the inner
diameter of the inlet pipe. By a combination of equations it follows for the equilibrium
loading
( )( )( )( ) dttyy
ytymMTp
ydwTpn
t
adsoutadsin
adsinadsout
bedadsds
adsingdsds ∫
−−
−=0 ,,
,,
tan
,2
tantan
11
14
πρ (4.15)
By continuous measuring of the mole fraction by a gas chromatograph the loading
can be calculated from the partial pressure. Schweighart [36] could reach low
pressure limits of 1.2 Pa with this method in adsorption systems like CO2-MS5A,
C2H4-MS5A, C2H6-MS5A and C3H8-MS5A.
4.3. Zero length column (ZLC) method
The zero length column technique was introduced in the beginning as a simple and
rapid approach for the study of sorption kinetics. If the same kind of experiment is
performed at a sufficiently low flow rate, the desorption rate will be determined by
convection under equilibrium conditions rather than by the desorption kinetics. The
effluent concentration history directly yields the equilibrium isotherm. The possibility
of using this approach to measure Henry constants was suggested by Ruthven et al.
[89,90].
In a ZLC experiment a couple of layers of zeolite is equilibrated in a short adsorber
with an adsorptive of known partial pressure. At time t = 0 the feed (carrier gas with
adsorptive) is switched to the pure carrier at constant flow rate. The concentration of
the desorbed adsorbate in the outlet stream of the ZLC is continuously measured.
The adsorber is so short that in can be treated as an ideal stirred tank reactor. The
justification for this is as follows: The two border cases for the residence time
behaviour of fluids in a pipe stream are the plug flow reactor without axial dispersion
and the ideal stirred tank reactor with a very large axial mixing. Characteristic for this
is the Bodenstein number Bo [93] with
68
axDLwBo = (4.16)
where Dax is the axial dispersion coefficient in the pipe, w the velocity of the fluid and
L the length of the pipe. With L → 0, as it occurs in a ZLC, the Bodenstein number
goes towards zero. This corresponds to the case of the ideal stirred tank reactor.
With this assumption it can be derived that the concentration of the outlet stream of
the ZLC after t = 0 is equal to the inner concentration of the adsorber. The differential
mass balance of the adsorptive is
( ) 0=++ tcVdtdcV
dtdnm GGads
& (4.17)
where VG is the gas volume in the adsorber, c the concentration of adsorptive in the
gas phase in mol/m3, GV& the gas stream in and t the time. Generally, the outlet
stream is not the same as the carrier gas stream and equation (4.10) is valid. The
adsorptive stream is equal to the third summand in equation (4.17):
( )tcVN Gads&& = (4.18)
With the mole fraction ( )ty adsout , and the overall number C of moles per m3 it is
( )tyCc adsout ,= (4.19)
Equation (4.10) can be rewritten in
( )( )( )tytyC
VNadsout
adsoutCads
,
,
1−= && (4.20)
The term dc/dt in equation (4.17) can be rewritten to
69
( )( ) ( ) ( ) ( )dt
tdyC
dttdy
CdtdCty
dttyCd
dtdc adsoutadsout
adsoutadsout ,,
,, =+== (4.20)
because in an isothermal system the overall number of moles per volume, C, does not
change which means that dC/dt = 0.
Equation (4.17) can be rewritten into the general form
( ) ( )( )( ) 0
1 ,
,, =−
++tytyC
Vdt
tdyCV
dtdnm
adsout
adsoutC
adsoutGads
& (4.21)
or
( )( )( )
( )ads
adsoutG
adsadsout
adsoutC m
tdyCV
mdt
tytyC
Vdn ,
,
,
1−
−−= & (4.22)
At t = ∞ it is yout,ads = 0 and n = 0. Now the start loading nt=0 can be calculated:
( )( )( ) ( )∫ ∫∫
∞=
==
−−
−=t
yadsout
ads
G
adsout
adsout
ads
C
n tadsoutt
tdymCV
dttyty
mCV
dn0
0
,,
,0
0,,01
& (4.23)
Therefore, it is
( )( )( )∫
∞=
== −−
=t
tadsoutads
G
adsout
adsout
ads
Ct y
mCV
dttyty
mCV
n0
0,,,
,0 1
& (4.24)
The loading n at time t can be calculated via equation (4.22) with
( )( )( ) ( )( )∫ == −−
−−=t
tadsoutadsoutads
G
adsout
adsout
ads
Ct yty
mCV
dttyty
mCV
nn0
0,,,,
,0 1
& (4.25)
With equation (4.24) we get for the loading n:
70
( )( )( )
( )( )( ) ( )∫∫ −
−−
−=
∞= t
adsoutads
G
adsout
adsout
ads
Ct
adsout
adsout
ads
C tymCV
dttyty
mCV
dttyty
mCV
n0
,,
,
0 ,
,
11
&& (4.26)
By integrating the adsorption isotherm in dependence of the measured concentration c
the loading n is obtained. This can be converted into partial pressure by means of the
law of ideal gases.
4.4. Concentration pulse chromatography method
The use of the concentration pulse chromatography for adsorbent screening is very
attractive since it is relatively inexpensive to set-up [91,92]. Further, this method is
capable of characterising adsorbents faster than other methods.
With the concentration pulse method, a pulse of sample is injected into the carrier
gas stream and passes through an adsorbent packed column. The response of the
column is measured as concentration vs. time at the exit of the column. From this
response peak a mean retention time of the sample, τ, is determined experimentally
with
( )
∫
∫∞
∞
−=
0
0
cdt
dttc Dττ (4.27)
The term τD is the mean dead time of the system. This dead time is the time required
for a sample pulse to pass the empty volume of the interconnecting tubing from
injection point to the detector and void space in the packed column.
With the that carrier gas composition the mean retention time is related to the
effective isotherm slope, K, by
( )
−+= K
wL
bed
bed
εετ 1
1 (4.28)
71
where εbed is the bed porosity. The dimensionless K factor can be transformed to the
dimensional Henry coefficient in mol/(kg*Pa) by
pelletTKHeρℜ
= (4.29)
where ρpellet is the apparent density of the adsorbent. With the former equations
(4.27) and (4.28) inserted in equation (4.29) it follows
( )
pelletadsbed
bedD
TLw
cdt
dttcHe
ρεε
τ
ℜ⋅
−⋅
−
−
=∫
∫∞
∞
11
1
0
0 (4.30)
The concentration pulse method is also applicable for determining binary adsorption
isotherms [91,92].
72
5. Prediction of Isotherms for p → 0 (Henry) The Henry coefficient is the first step in developing a pure component adsorption
isotherm that allows for overall prediction of the adsorption equilibrium. It is
characteristic for the region of diluted concentrations. In this work it is shown that the
Henry coefficient is not only dependent on the adsorption temperature but also on the
properties of the adsorptive and adsorbent themselves. It is a picture of the prevailing
interaction forces between the gas and solid molecules and ions. Understanding the
basic mechanisms of adsorption opens the door to an a priori approach of calculating
adsorption isotherms in a wide concentration range.
An adsorption isotherm in the diluted concentrations region is described by the Henry
law
pHen = (5.1)
where n is the loading in mol/kg, He the Henry coefficient in mol/(kg*Pa) and p the
equilibrium pressure in Pa. For graphical purposes this means that the loading n is
directly proportional to p with the slope He. All adsorption isotherms intersect with the
origin of the coordinate system. Logarithmising both sides of equation (5.1) leads to
equation
( ) ( ) ( )pHen logloglog += (5.2)
with slope 1 in a graph with double logarithmic axes. The intercept of the curve with
the log(n)-axes is equal to log(He) if the log(p)-axes begins at 1 Pa. Figure 5.1 shows
two different plots, one with linear and one with double logarithmic axes.
73
Figure 5.1: Two different views of adsorption isotherms in the low-pressure region.
Besides dimensionally correctness another important principle in chemical
engineering is the thermodynamic consistency of derivations. This goal is retained in
this work when the Henry coefficient is deduced from chemical potential equilibria. As
next step in this Chapter the Henry coefficient is calculated in the case of
energetically homogeneous adsorbents. An analytical function is derived from
theoretical considerations which is a powerful instrument for process engineers. The
last step introduces the general case of energetically heterogeneous adsorbents.
Adsorption isotherms in the low-pressure region
0
0,005
0,01
0,015
0,02
0,025
0 200 400 600 800 1000
Pressure p in Pa
Load
ing
n in
mol
/kg
Adsorption Pair 1Adsorption Pair 2Adsorption Pair 3
Adsorption isotherms in the low-pressure region
1E-06
1E-05
1E-04
1E-03
1E-02
1E-01
1E+00
1 10 100 1000
Pressure p in Pa
Load
ing
n in
mol
/kg
Adsorption Pair 1Adsorption Pair 2Adsorption Pair 3
74
5.1. Thermodynamic consistency
In Chapter 3 several descriptions of adsorption equilibria were introduced of which
some could be derived from thermodynamic consistent considerations. The Langmuir
isotherm is the most famous one. This work follows that principle.
Besides disperse forces mainly electrostatic forces influence the adsorptive during
exposion to an electrical field, e.g. the field of an ionic crystal lattice of a zeolite [8]. In
general these electrostatic forces contribute in a non negligible degree to the
adsorption equilibrium of an adsorptive with a solid surface. The overall adsorption
potential Φ can therefore be written as
∑∑ Φ+Φ+Φ+Φ+Φ+Φ+Φ=Φ=Φ −j
adsadsquadchaquadinddipchadipindindchaindindsolidgas (5.3)
The partial potentials are the integrated interaction energies of the adsorbate
molecule with the solid continuum structure or with other adsorbate molecules. They
describe the following interactions:
• The interaction of the induced dipole of the gas molecule with the induced
dipole of the solid molecule ( indindΦ ),
• the interaction of the induced dipole of the gas molecule with the electrical
charge of the solid molecule ( indchaΦ ),
• the interaction of the permanent dipole of the gas molecule with the induced
dipole of the solid molecule ( dipindΦ )
• the interaction of the permanent dipole of the gas molecule with the electrical
charge of the solid molecule ( dipchaΦ ),
• the interaction of the permanent quadrupole moment of the gas molecule with
the induced dipole of the solid molecule ( quadindΦ ) and
• the interaction of the permanent quadrupole of the gas molecule with the
electrical charge of the solid molecule ( quadchaΦ ).
75
In the following study for determining the Henry coefficient one important
assumptions is that the interactions between the adsorbate molecules are negligible.
The physical cause for this is the low concentration or partial pressure of the
adsorbate molecules in the Henry region, so that they do not interact with each other
but only with the solid surface. It is
0=Φ∑j
adsads (5.4)
A second assumption is that the solid elements do not have a permanent dipole or
quadrupole moment.
When two phases coexist in a system equilibrium is reached between them after a
certain time. In equilibrium state the chemical potentials µ of the two phases are
equal. For the adsorptive and adsorbate phases it follows that
adsorbateadsorptive µµ = (5.5)
The chemical potential is a function of pressure p and temperature T with
( )pTf ,=µ (5.6)
In the Henry region with its low pressure the adsorptive behaves like an ideal gas. Its
chemical potential is defined as
( ) ( )
+=
ref
adsadsrefadsadsorptiveadsorptive p
pkTpTpT ln,, µµ (5.7)
where pref is the reference pressure in Pa, ( )refadsadsorptive pT ,µ the reference chemical
potential of the adsorptive and k the Boltzman�s constant. The second term is the
correction from pressure pref to the adsorption equilibrium pressure of the gas pads.
The chemical potential of the adsorbate can be expressed as
76
( ) ( ) ∑ −Φ+
++= solidgas
ref
adsorbateadsadsrefadsadsorbateadsorbate p
dppkTpTpT ln,, µµ (5.8)
where ( )refadsadsorbate pT ,µ is the reference chemical potential of the adsorbate and
dpadsorbate is the difference pressure which acts additionally to the adsorbate layer.
The second term is therefore also a correction from the reference pressure to the
overall pressure of the adsorbate. The third term takes into account the sum of
interactions between the adsorbate and the solid. With the assumption
( ) ( )refadsadsorbaterefadsadsorptive pTpT ,, µµ = (5.9)
and that the adsorbate behaves like an ideal gas and according to
solid
adsadsorbate dv
dnTdp
ℜ= (5.10)
it follows with equations (5.7) and (5.8) that
∑ −Φ+
ℜ+
=
solidgas
ref
solid
adsads
adsref
adsads p
dvdnT
pkT
pp
kT lnln (5.11)
where dvsolid is the specific volume element of the solid. With algebraic conversions
and equation (5.1) it follows that
solid
v
v ads
solidgas
ads
dvkTTp
nHesolid
solid
∫ ∑=
−
−
Φ−ℜ
==0
1exp1 (5.12)
The specific volume element of the solid can be expressed with its specific surface
and the perpendicular distance z of the gas molecules from the surface:
dzSdv BETsolid = (5.13)
77
With equation (5.12) it follows the well known equation for the Henry coefficient
expanded with the overall interaction terms:
dzkTT
SHe
z
ads
solidgas
ads
BET ∫ ∑
−
Φ−ℜ
= −max
0
1exp (5.14)
The integration constant zmax depends on the structure of the solid. For massive
solids zmax is infinity, for porous solids it depends on the pore volume [47]. With
equations (5.3) and (5.4) the thermodynamic consistent equation for the Henry
coefficient becomes
( )dz
kTTS
Hez
ads
quadchaquadinddipchadipindindchaindind
ads
BET ∫
−
Φ+Φ+Φ+Φ+Φ+Φ−ℜ
=max
0
1exp (5.15)
5.2. Energetic homogeneous adsorbents
A special case of equation (5.15) is the adsorption of arbitrary gases on energetic
homogeneous adsorbents like active carbon or silicalite-1. For this case Maurer [26]
derived the potential function for the interaction energy indindΦ (z) which depends on
the critical properties of the adsorptive and the solid properties of the adsorbent. The
Henry coefficient reduces in this case to:
( )dz
kTz
TS
Hez
ads
indind
ads
BET ∫
−
Φ−ℜ
=max
0
1exp (5.16)
It is referred to the literature for the derivation of the interaction energy indindΦ (z) [26].
For consistency the main formulas shall be mentioned in this work with
( ) ( )3,
,
3
4
z
Cz
ji
jjiindind +
−=σ
ρπφ (5.17)
78
where Ci,j is an interaction constant between the adsorbate molecule and the solid
atoms, σi,j the minimum contact distance between the adsorption participants and ρj the
density of the solid. Figure 5.2 shows a principle scheme for the distance of the
adsorbate from the solid.
Figure 5.2: Principle scheme for the distance of the adsorbate from the solid [26].
The distance between the adsorbate sphere centre and the centre of the closest solid
atom in its vicinity is (σi,j + z). The minimum distance σi,j is calculated by the arithmetic
mean of the van-der-Waals diameters of the adsorbate and the solid molecules with
( )jjiiji ,,, 21 σσσ += (5.18)
Maurer [26] derived the van-der-Waals diameter of the adsorptive from its critical
properties. It is
3, 163
c
cii p
Tkπ
σ = (5.19)
where Tc and pc are the critical temperature and critical pressure, respectively. The
van-der-Waals diameter of the solid can be found in the literature. If not it is referred
to Section 5.2.2 for the derivation of the parameter σj,j.
j
σi,j + z
j
j
j j
j
j
j
j
i
79
The interaction constant itself can be formulated as
c
c
jj
j
Ai
jiji p
THaN
C 3,
3
3,
,
2169
σπρσℜ
= (5.20)
where Haj is the Hamaker constant of the solid.
5.2.1. The dimensionless Henry curve
The interaction energy indindΦ of two induced dipoles is also called the London
dispersion energy [5]. This dispersion energy can also be understood by the following
argument. Although the time-averaged dipole moment is zero for molecules without a
dipole, there exists at any time a finite value. An electromagnetic field emanates from
this dipole moment and induces a dipole in a nearby molecule. These dispersion
forces are omnipresent, they are always attractive.
The scope of this Section is to formulate a dimensionless function for the Henry curve
considering only London dispersion forces (Hamaker dominant). It is based on the
advancement of the Henry function formulated by equation (5.16) in the case of
energetic homogenous adsorbents. With equations (5.17) and (5.20) the interaction
energy indindΦ (z) can be rewritten by dividing by the energy unit (k.Tads)
( )( ) c
c
jj
j
Aj
ji
adsji
j
ads
indind
pTHa
NkTzkTz
3,
3
3,
3,
2169
3
4σπρ
σσ
ρπφ ℜ
+−= (5.21)
With
ANk=ℜ (5.22)
and the binomial formula
( ) 32233 33 bbabaaba +++=+ (5.23)
80
equation (5.21) can be simplified:
( )( ) ads
c
cjj
j
jijiji
ji
ads
indind
TT
pHa
zzzkTz
3,
32,
2,
3,
3, 2334
3σπσσσ
σφ+++
−= (5.24)
Therefore the Henry coefficient can then be expressed as
( ) dzTT
pHa
zzzTS
Hez
ads
c
cjj
j
jijiji
ji
ads
BET ∫
−
+++ℜ=
max
03,
32,
2,
3,
3, 1
2334
3exp
σπσσσσ
(5.25)
By dividing both sides by the minimum contact distance of the adsorbate molecule
and the solid atom σi,j equation (5.25) can be rearranged to
−
+
++
=ℜ
∫jiads
c
cjj
j
jijiji
jiBET
ads zdTT
pHa
zzzSTHe ji
z
,03,
3
,
2
,,
,
,max
12
3314
3expσσπ
σσσ
σ
σ
(5.26)
Both sides of equation (5.25) are now dimensionless. The left side is defined by
Mersmann et.al. [27] as the dimensionless Initial or Henry Loading IL with
jiBET
ads
STHe
IL,σ
ℜ≡ (5.27)
which is mainly a function of the Interaction Parameter IP with
cjj
j
ads
c
pHa
TT
IP 3,
2σπ
≡ (5.28)
and the integration variable
81
ji
zx
,
maxmax σ
≡ (5.29)
With the reduced distance
ji
zx,σ
≡ (5.30)
and the definitions
ILy ≡ (5.31)
and
IPa
34= (5.32)
the dimensionless Henry function can be expressed as
( ) ( ) dxxaxaxaa
dxIPxxx
ILxayxx
∫∫
−
+++
=
−
+++
==maxmax
032
032 1
331exp1
33143exp,
(5.33)
This integral cannot be solved analytically. It was numerically analysed with
commercially available mathematical programs. Nevertheless for given values of a
and xmax the integral can be calculated on a numerical basis for example with
MathCAD. In this context it is important to mention that the integration parameter a is
not a function of x. The constant a only depends on the solid and adsorptive
properties (the Hamaker constant Haj and the van-der-Waals diameter σj of the solid,
the critical temperature Tc and the critical pressure pc of the adsorptive) and also on
the adsorption conditions (the adsorption temperature Tads).
82
As next step the interval [0.1;30] of the dimensionless interaction parameter IP was
examined. This interval can be recalculated via equation (5.32) to an interval
[13.33;0.0444] of the integration parameter a. Through physical examinations a
minimum and maximum value for the reduced distance xmax was derived. For this the
following assumptions were made:
- The helium atom He has the smallest van-der-Waals diameter of gases which
is σHe = 2.66*10-10 m.
- The maximum pore diameter d of microporous solids is 2 nm = 20*10-10 m and
limits the distance of an adsorptive molecule from the solid surface.
- The van-der-Waals diameter of a solid atom is approximately 4.0*10-10 m.
- Large adsorptive molecules have a maximum van-der-Waals-diameter of
5.5*10-10 m.
- xmax is calculated via equation (5.34) to
( )jjii
ii
ji
dz
x,,
,
,
maxmax
21
2
σσ
σ
σ +
−== (5.34)
The values for xmax are then calculated as a minimum value of approximately 0.9 and
a maximum value of 2. With these values and by numerical calculations of the
integral (5.33) the table 5.1 was generated. Practically the integration values for the
dimensionless Henry loading IL do no longer differ above a value of 2 for the
interaction parameter IP. Both limits for the dimensionless Henry curve are shown in
figure 5.3. It can be seen that the Henry curve only depends on the interaction
parameter IP. It has to be mentioned that the curve does not cross the value pair
(0/1) which is shown by Mersmann et al. [27] as the thermodynamic consistent point
to be crossed.
For engineering purposes it would be easier to have an analytical function for
determining the Henry coefficient in the case of energetic homogenous adsorbents.
Therefore, in the next step an analytical function is introduced which is a powerful
and easy instrument for calculating the Henry coefficient.
83
Table 5.1: Value table for integral equation (5.33).
Dimensionless
parameter IP
Integration
parameter a
Integration
value IL for xmax
= 0.9
Integration
value IL for xmax
= 2
Integration
value IL for xmax
= ∞
0.1 13.3333333 0.028 0.034 0.0381 1.33333333 0.335 0.399 0.4412 0.66666667 0.953 0.986 1.075 0.26666667 5.383 5.752 5.966
10 0.13333333 101.933 102.822 103.26415 0.08888889 2632 2633 263420 0.06666667 80440 80440 8044025 0.05333333 2.67E+06 2.67E+06 2.67E+0630 0.04444444 9.34E+07 9.34E+07 9.34E+07
Figure 5.3: The dimensionless Henry curve � numerical values (equation (5.33)).
Dimensionsless Henry Curve
1E+00
1E+01
1E+02
1E+03
1E+04
1E+05
1E+06
1E+07
1E+08
0 10 20 30 40
Interaction Parameter IP=
Initi
al (o
r Hen
ry) L
oadi
ng IL
=(H
e ℜ
T)/(
σi,j
SB
ET)
with integral andxmax=0.9with integral andxmax=2
Cj
j
ads
C
pHa
TT
3
2πσ
84
The numerical values of figure 5.3 are approximated by the function
( )21exp pIPpIL = (5.35)
where p1 and p2 are parameters adjusted to the real numerical values of the integral
(5.33). The adjustment was performed by the numerical software Maple. For the two
cases of xmax = 0.9 and xmax = 2 the parameter values p1 = 0.3312917 and p2 =
1.1803227 for xmax = 0.9 and p1 = 0.3312936 and p2 = 1.1803211 for xmax = 2.0 were
obtained. The two parameter sets do not differ significantly. For this work the second
set was chosen. The exponential function (5.35) with the second set is shown and
compared with the numerical values in figure 5.4.
Figure 5.4: The dimensionless Henry curve � analytical regression.
Dimensionsless Henry Curve
1E+00
1E+01
1E+02
1E+03
1E+04
1E+05
1E+06
1E+07
1E+08
0 10 20 30 40
Interaction Parameter IP=
Initi
al (o
r Hen
ry) L
oadi
ng IL
=(H
e ℜ
T)/(
σi,j
SB
ET)
withintegralandxmax=2withregression
Cj
j
ads
C
pHa
TT
3
2πσ
85
The regression curve goes through the thermodynamic consistent point (0/1) [27].
The quality of the analytical form of the Henry curve is discussed in Chapter 7. It is
shown there also that this function does not cover only the adsorption behaviour of
energetic homogenous adsorbents like active carbon and silicalite-1 but also the
case of non-polar adsorbents paired with energetic heterogeneous adsorbents like
zeolites. This is a very new aspect derived for the first time in this work.
One of the assumptions in this Section was the value of the van-der-Waals diameter
of solids. The diameter for active carbon coal was chosen to be that of the carbon
atom [26], so σactive carbon = 3.40*10-10 m. In most cases there are no literature data for
this parameter. So in the next Section a way for determining this solid property is
derived.
5.2.2. The van-der-Waals diameter of solids
Because of the very strong dependency of the Henry curve on the interaction
parameter IP the van-der-Waals diameter of the solid has also a great significance. It
goes down as the third power in the calculation of the interaction parameter. With the
assumption that the Henry curve is thermodynamic consistent and characterised
accurately by equation (5.35) the following approach for the derivation of the
diameter σj of the solid can be made:
( ) ( ) ( )
==
+ℜ
=2
31
2exp
2p
cj
j
ads
cj
BETji
adsj p
HaTT
pgSTHe
fσπ
σσσ
σ (5.36)
It is:
( ) ( ) 0=− jj gf σσ (5.37)
where f and g are functions of the diameter σj as defined in equation (5.36). There is
only one unknown variable in equation (5.37), the diameter σj, if at least one
adsorption isotherm of the solid is known. For energetic homogenous adsorbents the
choice of the adsorptive is not critical. For energetic heterogeneous adsorbents the
86
adsorptive should have no dipole or quadrupole moment. Most of the solid�s surface
is examined by the BET-isotherm of nitrogen at 77 K as adsorption temperature. So
there is at least one experiment set of parameters (Henry coefficient, adsorption
temperature, critical properties and van-der-Waals diameter of the adsorptive,
specific surface and Hamaker constant of the solid) for each solid available without
further experimental work. In this study propane adsorption isotherms were taken to
calculate the van-der-Waals diameter of the examined solids. In the case of NaY
methane was taken. The zero point search as derived in equations (5.36) and (5.37)
was conducted with the program MathCAD. Table 5.2 shows the results.
Table 5.2: Van-der-Waals diameter of examined solids calculated with the
adsorptives propane and methane.
Solid, adsorptive
SBET (m2/kg)
Haj (J) TC (K) pC (Pa) Tads (K)
He (mol/kgPa)
σi (m) σj (m)
Active
carbon,
propane
1000000 6.0E-20 369.8 4.2E6 423 7.1E-5 4.15E-10 3.5E-10
Silicalite-1,
propane
372000 7.94E-20 369.8 4.25E6 303 1.06E-3 4.15E-10 3.8E-10
MS5A,
propane
727000 7.71E-20 369.8 4.25E6 343 5.36E-5 4.15E-10 3.8E-10
MS4A,
propane
650000 6.5E-20 369.8 4.25E6 423 4.22E-5 4.15E-10 3.8E-10
NaX,
propane
445000 8.25E-20 369.8 4.25E6 293 2.37E-3 4.15E-10 3.8E-10
NaY,
methane
500000 8.40E-20 190.4 4.60E6 343 1.16E-6 3,24E-10 4.2E-10
87
The results show that the van-der-Waals diameters of zeolites are in the same
magnitude as that of active carbon. Furthermore the chosen diameter for active
carbon of σactive carbon = 3.40 E-10 m is approved by this calculation by an accuracy of
+- 2.9 %.
5.3. Energetic heterogeneous adsorbents
In order to calculate the Henry coefficient for the adsorption of polar gases on
energetic heterogeneous adsorbents like zeolites it is necessary to take into account
all the possible interaction potentials between the adsorbate and the solid atoms. In
this Section analytical functions for these potentials are derived. Consequently the
Henry function of equation (5.15) can then be extended to an integral which is finally
numerically solvable. Later it will be shown that the interaction energy ads
indind
kTΦ is only
approximately 50 % of the sum of all interaction energies in the case of polar gases.
5.3.1. The range of polarities
In the following potential calculations the polarity properties of the adsorptives are
important factors. Their possible value intervals can be
• between 2.2*10-41 (C2*m2)/J (helium) and 2.13*10-39 (C2*m2)/J (n-decane) for
the polarisability iα of the gas molecule.
• The dipole moment µi can be between 0 (methane) and 1.0*10-29 C*m
(acetone).
• The quadrupole moment Qi can be between 0 (methane) and 6.21*10-39 C*m2
(anthracene).
5.3.2. Induced dipole gas � electrical charge solid interaction (indcha)
The assumption for the derivation of the interaction between an induced dipole of a gas
molecule and an electrical charge of a solid atom is that the solid is composed of so-
88
called average atoms with equal properties. In reality this is not true but it is effective
for the characterisation of the interaction energy.
The proposed route of van der Hoeven [41] for the calculation of the Hamaker constant
is based upon a subdivision of the solid molecules into atoms, which all have an
average composition and polarisability identical to that of the solid molecule. The
constants (London constant, polarisability, number density) are then calculated from
these average atomic properties. In this approach it is further supposed that the atoms
are all bonded by semi-polar bonds and that only the bond-forming valence electrons
contribute to the disperse energetic waves with respect to the Hamaker constant.
According to Münstermann [131] the adsorption of polar or polarisable molecules is
preferred by molecular sieves. These adsorbents are attracted by the positive charged
cations of the zeolite cages. This effect has to be taken into account when dealing with
charge interactions for calculating the Henry coefficient of arbitrary gases.
The potential energy Eindcha between two molecules or atoms, of which one is
polarisible and the other one possesses an electrical charge, can be described by
Hirschfelder [19]
( ) ( )20
4
2
421
r
jiindcha
r
qrE
επε
α−= (5.38a)
where iα is the polarisability of the gas molecule in (C2*m2)/J, qj the effective charge of
the average atom of the solid in C, ε0 the permittivity of free space = 8.854*10-12
C/(V*m), εr the relative permittivity of the adsorbent and r the distance between the
centre of the molecules in m. For calculation of the relative permittivity of microporous
adsorbents the porosity of the solid has to be taken into account. It is assumed that
( ) 211 nr ⋅−+⋅= ϕϕε (5.38b)
where ϕ is the porosity and n the refractive index of the solid. For MS5A the relative
permittivity is 1.73 with ϕ = 0.5 [24] and n = 1.57.
89
From a summation of all pair potentials it follows the interaction potential of the gas
molecule with the whole solid continuum. In general it is [21]
( ) drrrE atj2
, 4πρσ∫∞
=Φ (5.39)
where ρj,at is the number density of solid atoms in 1/m3 calculated with equation (3.92).
In this work the following general description of the interaction potentials was chosen:
( ) drrrE atjzji
2, 4
,
πρσ∫∞
+
=Φ (5.40)
The lower integration limit of equation (5.39) was substituted by (σi,j + z) because the
minimum distance between the molecules is σi,j, also known as the minimum van-der-
Waals contact distance. The effective mean charge of the average solid atom qj,at
depends on the average number of positive charged cations scations per solid atom. This
property is introduced here for the first time as
a
ccations n
ns = (5.41a)
where nc are the number of positive charges and na the number of atoms in the solid
molecule. For example for MS5A nc = 12, na = 79 and scations = 0.152.
It follows for the effective mean charge that
esq cationsj = (5.41b)
where e is the charge of one electron (e = 1,60*10-19 C). The interaction energy
( )zindchaΦ is then due to equations (5.41b) and (5.38a)
( ) ( ) ( )jir
atjcationsiatj
z r
jiindcha z
esdrr
r
qz
ji ,22
0
,22
2,2
04
2 142
1442
1
,σεπε
ραπρ
επε
α
σ +−=−=Φ ∫
∞
+
(5.42a)
90
With the same substitution as in equation (5.30) and after dividing both sides of
equation (5.42a) by the energy unit (k*Tads) one gets the reduced interaction energy
( )( ) ( )1
1181
,22
0
,22
+−=
ΦxTk
eskT
x
jiadsr
atjcationsi
ads
indcha
σεπερα
(5.42b)
The dimensionless interaction parameter for the actual case is then defined as
( ) jiadsr
atjcationsiindcha Tk
esA
,22
0
,22
8 σεπερα
= (5.43)
5.3.3. Permanent dipole gas � induced dipole solid interaction (dipind)
Gas molecules can have a permanent dipole when the atoms of different charge of the
molecule compound are not arranged symmetrically to each other. For example,
trifluormethane CHF3 has a dipole moment of 1.6 debye or 5.344*10-30 C*m.
Permanent dipoles induce a dipole in the attraction partner if this partner is polarisible.
The potential attraction energy between two molecules is according to Debye [4]
( )( )2
06
2
4 r
jidipind r
rEεεπ
αµ−= (5.44)
where µi is the dipole moment of the gas molecule in Cm and jα the polarisability of a
solid atom according to equation (4.49) in (C2*m2)/J. In the following it is assumed that
the solid atoms are induced from the polar gas molecules but do not have a permanent
dipole themselves. The interaction energy of one polar gas molecule with the whole
solid continuum is with equation (5.40)
( ) ( ) drrrEz atjzdipinddipind
ji
2, 4
,
πρσ∫∞
+
=Φ (5.45)
Therefore it is
91
( ) ( )3,
220
,2 1
12 zz
jir
atjjidipind +⋅
−=Φσεπε
ραµ (5.46)
After introduction of the reduced location variable x the reduced interaction energy
permanent dipole gas � induced dipole solid is
( )( )33
,22
0
,2
111
12 xkTkTx
jiadsr
atjji
ads
dipind
+⋅−=
Φσεπε
ραµ (5.47)
The dimensionless interaction parameter Adipind is therefore
( ) 3,
220
,2
12 jiadsr
atjjidipind kTA
σεπεραµ
⋅= (5.48)
5.3.4. Permanent dipole gas � electrical charge solid interaction (dipcha)
This interaction type is dominant for highly polar adsorptives like trifluormethane
(CHF3), ammonia (NH3) and methanol (CH3OH). The potential attractive energy of a
permanent dipole with a charged atom is quantified [19] with
( )( )2
04
22
431
r
ji
adsdipcha r
qkT
rEεπε
µ−= (5.49)
with already known parameters. The summation of the potential energy over the whole
solid yields
( ) ( )∫∞
+
=Φijz
atjdipchadipcha drrrEzσ
πρ 2, 4 (5.50)
After integration and deviding by the energy unit (kTads) and introducing the
dimensionless variable x it follows for this interaction energy
92
( )( ) ( )xkT
eskT
x
jiadsr
atjcationsi
ads
dipcha
+⋅−=
Φ
111
12 ,222
0
,222
σεπερµ
(5.51)
Then the dimensionless interaction parameter Adipcha can be defined as
( ) jiadsr
atjcationsidipcha kT
esA
,222
0
,222
12 σεπερµ
⋅= (5.52)
5.3.5. Quadrupole moment gas - induced dipole solid interaction (quadind)
The quadrupole moment is a fundamental property for the description of the molecular
charge distribution [103]. In the theory of electricity there are two signs of electric
charge. The net total charge is the monopole moment. If there are charges of two
signs separated, then there exists a dipole moment along the line connecting the
charges. If there are charges of both signs, but separated in a more complicated way,
an electric quadrupole may be present. The quadrupole moment gives an indication of
the derivation of the charge distribution of the adsorptive from spherical symmetry.
For gases with strong quadrupole moments like CO2 the previous potential
phenomena are not sufficient to describe their great adsorption effect in the Henry
region particularly because their permanent dipole is small or even zero in the case of
benzene. The cause is that the quadrupole moment can also induce a dipole moment
on the solid atoms or can interact with electrical charges of the adsorbent. For two
molecules of which one has a quadrupole moment and the other is polarisible the
potential energy is [19]
( )( ) 82
0
2
423
r
QrE
r
jiquadind επε
α−= (5.53)
where Qi is the quadrupole moment of the adsorptive in C*m2. It follows for the
reduced interaction potential with the usual procedure as in the previous chapters that
( )( )55
,22
0
,2
111
403
xTkQ
kTx
jiadsr
atjji
ads
quadind
+−=
Φσεεπ
ρα (5.54)
93
Now the dimensionless interaction parameter Aquadind can be directly specified as
( ) 5,
220
,2
403
jiadsr
atjjiquadind Tk
QA
σεπερα
= (5.55)
5.3.6. Quadrupole moment gas � electrical charge solid interaction (quadcha)
This potential is in the same order of magnitude as the permanent dipole � electrical
charge interaction. The potential energy between two molecules with a quadrupole
moment and an electrical charge is according to Israelachvili [21]
( )( ) 62
0
22
4201
r
qQkT
rEr
ji
adsquadcha επε
−= (5.56)
It follows for the reduced interaction potential:
( )( ) ( )33
,222
0
,222
111
2401
xkT
esQkT
x
jiadsr
atjcationsi
ads
quadcha
+−=
Φσεπε
ρ (5.57)
with the well known parameters of the adsorptive and the adsorbent. The
dimensionless interaction parameter Aquadcha is then
( ) 3,
2220
,222
240 jiadsr
atjcationsiquadcha kT
esQA
σεπερ
= (5.58)
Later in chapter 7.2 the validity of this theoretical model according to the equations
(5.24), (5.42b), (5.47), (5.51), (5.54) and (5.57) will be discussed.
5.3.7. Overall adsorption potential
The overall adsorption potential between a gas molecule and the solid continuum can
be expressed as the sum of the partial potentials as shown in equation (5.3). Because
94
the Henry coefficient depends on the reduced overall potential ads
SolidGas
kT∑ −Φ−
as shown
in equation (5.14) also the partial potentials will be expressed by reduced equations
relative to the energetic unit kTads. Therefore, the induced dipole � induced dipole
interaction can be rewritten as:
( )( ) ads
c
cjj
j
ji
ji
ads
indind
TT
pHa
zkTz
3,
3,
3, 2
4
3σπσ
σ+
−=Φ (5.59)
It is assumed that this potential is the London potential expressed by a function of the
Hamaker constant of the solid and the critical parameters of the adsorptive. This item
will be discussed in Chapter 7. With the additional equations (5.41b), (5.46), (5.50),
(5.54) and (5.57) for the other partial potentials the reduced overall potential can be
formulated in detail as shown in equation (5.60).
( ) ( )
( ) ( ) ( ) ( ) ( ) ( )
( ) ( )
( ) ( ) ( )
( ) ( ) ( )3,
2220
,222
5,
220
,2
,222
0
,222
3,
220
,2
,22
0
,22
3,
3,
3,
1240
140
3
112
112
18
2
4
3
zkT
esQ
zTkQ
zkT
es
zkT
zkTes
TT
pHa
z
kTzzzzzz
kTz
kTz
jiadsr
atjcationsi
jiadsr
atjji
jiadsr
atjcationsi
jiadsr
atjji
jirads
atjcationsi
ads
c
cjj
j
ji
ji
ads
quadchaquadinddipchadipindindchaindind
ads
solidgas
ads
++
++
++⋅
++⋅
+
++
++
=
=Φ−Φ−Φ−Φ−Φ−Φ−
=
=Φ−
=Φ− ∑ −
σεπερ
σεεπρα
σεπερµ
σεπεραµ
σεπερα
σπσσ
(5.60)
In figure 5.5 the interactions of the adsorptive molecule with the solid continuum are
presented graphically. The arrows are numbered respectively to the partial potential
energies which occur between adsorptive and adsorbent. It has to be mentioned that
the polarisability αi is the electronic polarisability of the molecule according to
95
Israelachvili [21] for further theoretic considerations. (Most important in figure 5.5 are
the potentials indind, dipcha and quadcha, compare Section 7.3.5).
96
Figure 5.5: The partial potentials between adsorptive and adsorbent.
Gas i Solid j
induced dipole (ind)
permanent dipole (dip)
quadrupole (quad)
α i
µ i
Q i
α j
q j
induced dipole (ind)
charge (cha)
1
2
3
4
5
6
(1) indind
(2) indcha
(3) dipind
(4) dipcha
(5) quadind
(6) quadcha
( )( )
( )( )
( )( )
( )( ) ( )
( )( )
( )( ) ( )3
,222
0
,222
5,
220
,2
,222
0
,222
3,
220
,2
,22
0
,22
3,
3,
3,
1240
140
3
112
112
18
243
zkT
esQkT
z
zTkQ
kTz
zkT
eskT
z
zkTkTz
zkTes
kTz
zTT
pHa
kTz
jiadsr
atjcationsi
ads
quadcha
jiadsr
atjji
ads
quadind
jiadsr
atjcationsi
ads
dipcha
jiadsr
atjji
ads
dipind
jiadsr
atjcationsi
ads
indcha
ji
ji
ads
c
cjj
j
ads
indind
+=
Φ−
+=
Φ−
+=
Φ−
+=
Φ−
+=
Φ−
+=
Φ−
σεπερ
σεεπρα
σεπερµ
σεπεραµ
σεπερα
σσ
σπ (5.59)
(5.42b)
(5.47)
(5.54)
(5.57)
(5.51)
97
5.3.8. Henry coefficient for energetic heterogeneous adsorbents
By inserting the interaction parameters into equation (5.15) the dimensionless Henry
coefficient becomes
( ) ( ) ( )∫
−
++
+
+++
++
=max
053 1
111exp
xquadindquadchadipindindindindchadipcha dxx
A
x
AAAxAA
IL (5.61)
where Aindind is defined as the interaction parameter for induced dipole � induced dipole
interaction
Ads
C
j
jindind T
Tpc
HaA 3
243
πσ= (5.62)
The integral increases very strongly with increasing distance x, for small values of x it
reaches a limit. In this work it is recommended to use a limiting value of 2 for x. The
integral itself is not analytically solvable. It is not possible to introduce a regression
function because it is a three-dimensional problem. Nevertheless, with commercial
software like MathCAD numerical values for the Henry Loading IL can be evaluated.
This value can be transformed to the Henry coefficient due to
ads
BETji
TSIL
He⋅ℜ
⋅⋅= ,σ
(5.63)
Therefore, it is possible to compare the theoretical a priori calculated Henry coefficient
with experimental data. The value pairs should lie on a diagonal in a coordinate system
with the experimental Henry values as the x-axes and the calculated ones as the y-
axes. The quality of the model will be discussed in Chapter 7.
98
5.4. Calculation methods for the Henry coefficient
5.4.1. Henry coefficients from experiments and literature
The Henry coefficients from experiments were derived in this work by applying the
Langmuir and Toth formulae for adsorption, in this study the equations (3.109) and
(3.119).
The proposed Langmuir method is a graphical method from which the Henry
coefficient can be obtained by regressing the transformed data by a linear curve. The
intercept of this linear curve with the y-axis is indirectly proportional to the Henry
coefficient. In this plot the y-axis shows the ratio of the equilibrium pressure p to the
equilibrium loading n in Pa/(mol*kg) whereas the x-axis shows the equilibrium
pressure p itself in Pa. The data were analysed by using commercial programs.
The same experimental data have also been analysed by the Toth equation (3.119)
due to an analytical fitting method because a three parameter isotherm cannot be
represented by a two dimensional graph. Model parameters are obtained using the
non-linear least square method of Margules. The parameters are optimised based on
the minimum sum of square error given by
( )∑ ∑= =
−+
=
NP
i
NP
i
nnnn
ss1 1
2experimentmodel
2
experiment
model (5.64)
where ss is the sum of square errors and NP the number of data points from the
experiment.
In many literature essays Henry coefficients are tabulated or given as a function of
temperature and heat of adsorption. If not, adsorption isotherms were analysed by the
two mentioned methods of Langmuir and Toth. In some of the essays the authors did
not tabulate the data points of equilibrium pressure and loading. In such cases the
adsorption graphs were scanned and the data points were quantified by the graphical
program Origin .
99
5.4.2. Theoretical Henry coefficients
Theoretical Henry coefficients have been calculated by using the derived equations
(5.59) and (5.61) in the commercially available mathematic program MathCAD
(Version 13, Mathsoft Engineering & Education, Inc.). This program has a Maple kernel
to calculate mathematical equations. Equation (5.65) was used in MathCAD to
calculate the dimensionless Henry coefficient of Z different adsorption pairs with
(5.65)
where the interaction parameters Ai,j are one-dimensional vectors with length of Z and
x is the reduced distance as in equation (5.30). The result IL, also a vector with length
of Z, is written back to a calculation sheet. The unknown coefficients fi,,j have been
determined by theoretical and experimental considerations and will be discussed in
Chapter 7. They can be interpreted as potential factors of the interaction parameters
Ai,j. At a first approach it is expected that these potential factors should be close to unity
or even fi,j = 1. However, experimental results show that in the case for interactions
which refer to the electrical charge q derived in the Sectionss 5.3.2, 5.3.4 and 5.3.6 the
factors are only close to unity. All other factors are equal to 1. The reasons for this
behaviour may be the fact that the distance of the different atoms and their angles with
respect to the dipole axis of the adsorptive molecule differs from atom to atom.
Therefore all average values depend on these distances and angles. For instance,
equation (5.49) is valid for freely rotating molecules. This may not be true for molecules
adsorbed in the cage of a zeolite. Israelachvili [21] has shown that calculations with the
objective to average distances and angles are problematic. It is assumed that these
average values can be calculated by complex integrals. In Chapter 7 the fi,,j are derived
from theoretical considerations and experimental results. They will be discussed later
in more detail.
IL
vi
0
2
xexp
A_QUADIND i
f_quadind
1 x+( )5
0.75A_INDIND i
f_indind⋅
A_DIPIND i
f_dipind+
A_QUADCHA i
f_quadcha+
1 x+( )3+
A_DIPCHA i
f_dipcha
A_INDCHA i
f_indcha+
1 x++
1−
⌠⌡
d←
i 0 Z..∈for
vreturn
:=
f_quadind
100
6. Prediction of Isotherms for p →∞ (Saturation) The a priori prediction of adsorption isotherms based only on the properties of the
adsorptive and the adsorbent molecules is a challenging goal. In this Chapter a new
model will be introduced which enables the prediction of the isotherm from the Henry
to the saturation region for Hamaker dominant adsorption pairs and for systems
where the adsorptive is polar and the adsorbent is heterogeneous with respect to
energy.
Examples for the Hamaker dominant case are the adsorption of an unpolar gas like
an alkane on active carbon or any molecular sieve or the adsorption of polar
adsorptives on any energetic homogenous adsorbent. Furthermore the fourth case is
the adsorption of any polar molecule like methanol on energetically heterogeneous
microporous solids like zeolite NaY.
The idea of the presented new model is that any isotherm of type I has two boundary
conditions for the two variables loading n and partial pressure p. The first boundary
condition is that in the Henry region the loading and the equilibrium pressure have a
constant ratio, equal to the Henry coefficient. The second boundary condition is that
in the saturation region the slope of the curve n=f(p) becomes zero. With a differential
equation for the variables the intermediate range it is possible to calculate the whole
adsorption isotherm mathematically.
In this model it is assumed that a narrow relationship between the relevant interaction
energies and the Henry coefficient exists. The saturation is calculated only in
dependence on the properties of the adsorptive and the adsorbent.
In Chapter 7 the quality of this new model will be validated by comparing the
predicted isotherms with data from the literature and also with data from own
adsorption experiments.
101
6.1. Differential form
The maximum loading n∞ for p → ∞ is given when the micropore volume is
completely filled with adsorbate which has the molar volume β(T):
( )Tv
Mv
n mciro
i
adsorbmicro
βρ
==∞ (6.1)
The relationship between
• the pore filling degree ∞
=nnns
*
• the number N of molecule layers
• the microporosity parameter micro
jBET
vS σ
and
• the adsorbate parameter ( )TN iA
βσ 3
can be described by
( )3
*
iAmicro
jBETs N
TvS
Nnnn
σβσ
⋅⋅==∞
(6.2)
The microporosity parameters of the adsorbents investigated in this work are in the
range 0.60 < micro
jBET
vS σ
< 1, see as examples tables 7.4, 7.5 and 7.6. Nearly all
adsorptives with molar masses above M = 16 kg/kmol valid for methane have
adsorbate parameters in the range 0.5 < ( )TN iA
βσ 3
< 0,75 at a temperature of 300 K, see
as examples tables 7.4, 7.5 and 7.6. This means that the maximum number Nmax of
102
molecular layers is between Nmax = 0.3 up to Nmax = 1 for the pore filling 1* ==∞nnns
which corresponds for a complete filling of all micropores.
According to Mersmann [122] the Henry range of an adsorption isotherm is finished
when the number of layers, NHe, assumes approximately the value 0.1. This means
that the borderline between the Henry range and the transition range can be
expected between Hen
n
∞
= 0.1 and Hen
n
∞
= 0.2. In the loading range
Henn
nn
<
∞∞
the adsorption places with the highest energies will be occupied.
Therefore, the interaction energy is smaller than the overall energy ( )adskTzΦ− in
equation (5.60) in the case Hen
nnn
>
∞∞
. It can be shown that the energy ads
indind
kTΦ
−
is at least 50 % up to 93 % of the overall energy.
It is further assumed that in the transition range between the Henry and the saturation
region the Hamaker energy is dominant in comparison to other possible potentials.
According to Braun [4] and Hirschfelder et. al. [19] Hamaker potentials are long-range
interactions. For the first layer of adsorbate molecules on the solid surface the
electrostatic potentials have to be taken into account as done for the calculating of the
Henry coefficient. In the second layer the Hamaker energy is dominant because of its
long-range attraction compared to the electrostatic forces.
Furthermore, the reduction of the interaction energy is caused by the presence of
adsorbate molecules and by their distance from the surface molecules of the
adsorbent. According to equation (5.24) the energy is reduced by a factor 3
21
=
0.125 for adsorptive molecules in the second layer. Therefore, the slope )(ln)(lnpdnd of
an isotherm decreases with an increasing pore filling
∞nn and increasing interaction
energy. This is demonstrated in figure 6.1 where the pore filling of some adsorptives
103
is plotted against the pressure. The higher the interaction energy the smaller is the
pressure valid for leaving the Henry range and entering the transition range.
Figure 6.1: Adsorption isotherms on active carbon coal, fitted with Langmuir
(methane/ac and ethane/ac) and Toth (acetone/ac and methanol/ac).
Consequently, adsorptives with high 2
3
3
cj
j
ads
c
pHa
TT
σ data lead to a wide transition
range over several orders of magnitude of the pressure and their mean slope )(ln)(lnpdnd
in the transition range is small. This slope assumes zero for 1→∗sn (saturation). All
this ideas lead to the following differential equation which fulfils these requirements
[122]:
( )
−+
==
∞
∞∗
nnnn
ETCnp
dpdn
pdnd
Haadsr1
1
1)(ln)(ln
23
_
(6.3)
with ( )( ) 1lnln =pdnd for n → 0 and ( )
( ) 0lnln =pdnd for n → n∞.
1E-04
1E-03
1E-02
1E-01
1E+00
1E+00 1E+01 1E+02 1E+03 1E+04 1E+05 1E+06 1E+07 1E+08
Adsorption pressure p in Pa
Pore
filli
ng d
egre
e
methane/ac301K
ethane/ac303K
methanol/ac298Kacetone/ac
298K ( )( )
( )( )
( )( )
434210
*
lnln
lnln
lnln
≈
∞−=pdnd
pdnd
pdnd Sslope:
Henry border line
104
It is assumed that C is a constant, and its order of magnitude can be evaluated from
the condition that the expression ( )
−∞
∞∗
nnnn
ETC Haadsr1
23
_ must be noticeable
compared to 1 (say 0.1) when leaving the Henry range. In this work C = 0.55 * 10-3.
Some parameters in equation (6.3) are dimensionless numbers and can be defined
as
c
adsadsr T
TT =_ (6.4a)
cj
jHa p
HaE 3
*
σ= (6.4b)
where Tr_ads is the reduced adsorption temperature and EHa* the reduced Hamaker
energy. The loading n∞ is equal to the saturation loading when the micropore volume
vmicro is completely filled with adsorbate of the density ρadsorb and can be calculated by
equation (6.1) where vmicro is in m3/kgadsorbent, the density ρadsorb in kgadsorbate/m3 and
the molar mass Mi in kgadsorbate/moladsorbate.
The calculation of the adsorbate density, especially in the supercritical region, is
afflicted with uncertainties because the physical state of an adsorbed molecule
cannot be described definitely. In this work, if no data for the density of the liquid was
available, the ratio of molar mass of the adsorbate and adsorbate density is
calculated by the molar volume of the adsorbate β [36] which depends on the
adsorption temperature Tads. The molar volume is inter- and extrapolated between
the molar volume of the adsorptive molecule at the normal boiling point vb and the
van-der-Waals volume vvdW [36]. It is
( ) ( )bvdWbc
badsbads
adsorb
i vvTTTT
vTM
−
−−
+== βρ
(6.5)
105
where Tb is the normal boiling temperature of the adsorptive. Tb and the normal molar
volume of the adsorptive vb are tabulated for many cases in the literature [106]. For
any other case the normal molar volume in m3/mol can be calculated by the method
of Tyn and Calus [106] with
048.16 *10*285.0 cb vv −= (6.6)
where vc is the critical volume of the adsorptive in cm3/mol, see figure 6.2.
Figure 6.2: Normal molar volume of the adsorptive in dependence of the critical
volume [106].
The van-der-Waals volume vvdW can be expressed by equation (3.29). The saturation
loading n∞ is now
( )ads
micro
Tv
nβ
=∞ (6.7)
The ratio of the loading n to the saturation loading n∞ is defined as the saturation
degree ns* with
adsorbmicro
is v
Mnnnn
ρ==
∞
* or (6.8)
Method of Tyn and Calus [106]
020406080
100120140160180200
0 100 200 300 400 500
Critical volume of the adsorptive vc in cm3/mol
Nor
mal
mol
ar v
olum
e of
the
adso
rptiv
e v
b in
m3 /m
ol
106
( )micro
s vTn
nnn β==∞
* (6.9)
Equation (6.3) can be rewritten with the definition
23
3
=
cj
j
ads
c
pHa
TT
CBσ
(6.10)
which is independent of the loading n and the pressure p to
)(ln1
1)(ln pdnnnn
Bnd =
−+
∞
∞ (6.11)
6.2. Integrated form with boundary conditions
Generally, three cases can be distinguished with respect to the slope ( )( )pdnd
lnln of an
adsorption isotherm:
• ( )( ) 1lnln =pdnd (dilute or Henry range) (6.12)
• ( )( ) 0lnln =pdnd (saturation range) and (6.13)
• ( )( ) ( )( )∗= SHaadsrs nETnfgpdnd ,,,
lnln *
_* (6.14)
6.2.1. The dilute region
The first case is the region where the loading n is much smaller than the saturation
loading n∞, so
∞<< nn (6.15)
107
The differential equation (6.11) reduces to equation (6.12). The integration of the
differential equation (6.12) leads to
( ) ( )∫∫ = pdnd lnln or (6.16)
.lnlnln Constpn += (6.17)
The integration constant Const. can be interpreted as the Henry coefficient:
HeConst ln.ln = (6.18)
It follows that
0ln =
pHen or (6.19)
pHen = (6.20)
which is known as Henry�s law. Some authors doubt that a real Henry region exists
with respect to the more or less degree of heterogeneity of the adsorbent surface.
However, dealing with chemical engineering predictions the assumption of a Henry
range is reasonable and helpful.
6.2.2. The saturation region
In the saturation region the slope of the loading n relative to the equilibrium pressure
does not change significantly. The limit of the differential equation (6.3) is for n → n∞
0
11
1lim
11
1lim)(ln)(lnlim
23
3
=
−+
=
−
+
=
∞
∞
→
∞
∞
→→ ∞∞∞
nnnn
Bnnnn
pHa
TT
Cpdnd
nn
cj
j
ads
c
nnnn
σ
(6.21)
108
or
( ) 0ln =nd or (6.22)
∞== nconstn . (6.23)
6.2.3. The general case
In the regime between the Henry and the saturation region the differential equation
(6.11) has to be solved. It follows with
( )ndnnd =ln (6.24)
the integral form
( ) ( )∫ ∫ ∫=−
+
∞
∞ pdndn
nnnn
Bnd ln1
ln or (6.25)
HepnnBn lnln1lnln +=
−−
∞ (6.26)
Equation (6.26) can be rewritten to
B
nn
nnB
npHe
−
∞∞
−=
−−=
1ln1lnln (6.27)
It follows the adsorption isotherm description for the general case (see figures 6.3
and 6.4) with
109
B
nnn
pHe
−=
∞
1
1 or (6.28):
B
nnHe
np
−=
∞
1
1 (6.29)
110
Figure 6.3: Pore filling for the general case.
1E-08
1E-07
1E-06
1E-05
1E-04
1E-03
1E-02
1E-01
1E+00
0 0,2 0,4 0,6 0,8 1
Pore Filling
B = 2
B = 3
B = 5
B = 7
B = 10
23
3
=
cj
j
ads
c
pHa
TTCB
σ
p He n
∞nn
111
Figure 6.4: Residual pore filling for the general case
1E-05
1E-04
1E-03
1E-02
1E-01
1E+00
0,0010,010,11
Residual Pore Filling
B = 2
B = 3
B = 5
B = 7
B = 10
p He n
23
3
=
cj
j
ads
c
pHa
TT
CBσ
∞
−nn1
112
7. Results and Discussion
7.1. New experimental results
The experiments carried out for measurements are described in the appendix B2.
Experiments were carried out in Garching and Leipzig.
7.1.1. Adsorption isotherms
Adsorption isotherms of methanol and dichloromethane were measured on zeolites
MS5A, NaX and NaY. These systems were chosen to expand literature data with
experimental data in the case of adsorption of polar molecules or such with high
polarisabilities on energetic heterogeneous sorbents. The experiments were carried
out in the pressure range from 0.1 Pa to saturation pressure of the adsorptive at room
temperature. Figure 7.1 shows adsorption isotherms of methanol on several zeolites.
Figure 7.2 shows adsorption isotherms of dichloromethane on the same zeolites. The
temperatures ranged from 44°C to 47°C.
As can be seen in figure 7.1 methanol has the highest adsorption affinity with NaY
followed by MS5A and NaX, which seem to have the same adsorption attraction to
methanol. NaY was pure crystalline, MS5A and NaX have been used as industrial
pellets with binder amount. The MS5A zeolite was grated to powder, pressed
together and then broken to small pieces for the glass crucible. Adsorption was
repeated for this modified molecular sieve. The experiment shows that there is some
diffusion resistance when adsorbing into industrial pellets with binder amount. For the
theoretical calculations only the data of the crystals and grated pellets were taken.
Reliable data from the Garching experiments begin at 0.1 Pa, the reliable data from
Leipzig begin at 10 to 100 Pa. The saturation loading for methanol on the examined
zeolites was in the magnitude of 6 to 6.5 mol/kg. This is in good agreement with the
theory of micropore filling as will be shown later. The Henry region of methanol
adsorption on zeolites ends at high loadings of approximately 3 mol/kg or 10 Pa,
113
respectively. In the experiments the Henry region could be reached very well down to
pressures of 0.1 Pa.
Figure 7.1: Methanol adsorption on zeolites NaY, MS5A and NaX.
Dichloromethane adsorption was conducted with crystal NaY and grated MS5A
zeolite pellets and additionally with NaX zeolite. The Henry region ends at
approximately 2 Pa for MS5A and 10 Pa for NaY. The saturation loading is in the
range of approximately 3.5 mol/kg, less than methanol because of the higher molar
mass of approximately 84 kg/kmol. Henry coefficient calculations will be compared
later regarding to Langmuir and Toth regressions.
With this experiments it could be shown that the Henry region and the saturation
region of solvents with moderate dipole moments of approximately 1.5 debye on
zeolites can be measured with gravimetric adsorption methods.
0,01
0,1
1
10
0,1 1 10 100 1000 10000
Pressure p in Pa
Load
ing
n in
mol
/kg
CH3OH on NaY, crystal, 318.4 K
CH3OH on MS5A, grated pellets, 319.35
CH3OH on MS5A, pellet, 317.35 K
CH3OH on NaX, pellet, 319.4 K
114
Figure 7.2: Dichloromethane adsorption on zeolites NaY, MS5A and NaX.
7.1.2. Comparison of Langmuir and Toth regression quality
In this Section the experimental data are fitted with both Langmuir and Toth
adsorption isotherms. Figure 7.3 shows the adsorption of methanol on NaY and in
figure 7.4 the adsorption equilibrium of dichloromethane on MS5A zeolite is shown. In
both cases the Toth isotherm delivers better fits to the experimental data than the
Langmuir isotherm. The standard deviation is 8.1 % for Toth and 17.3 % for Langmuir
in the case of methanol adsorption on NaY. Furthermore the standard deviation is 3.8
% for Toth and 5.8 % for Langmuir in the case of dichloromethane adsorption on
MS5A. In Table 7.1 the standard deviations for the experimental data are listed. On
the basis of this table it was decided to calculate the Henry coefficient with Toth
regression which fitted better to the data. Toth parameters nmax, KT and m and
Langmuir parameters nmon and KL are explained in Chapter 3.
0,01
0,1
1
10
0,1 1 10 100 1000 10000
Pressure p in Pa
Load
ing
n in
mol
/kg
CH2Cl2 on NaY, crystal, 318.4 K
CH2Cl2 on MS5A, grated pellets, 319.4K
CH2Cl2 on NaX, pellet, 316.9 K
115
Figure 7.3: Methanol (CH3OH) adsorption on crystal NaY at 318.4 K.
Figure 7.4: Dichloromethane (CH2Cl2) adsorption on MS5A, grated pellets at 319.4 K.
Adsorption of methanol on crystal NaY at 318.4 K
0,01
0,1
1
10
0,1 1 10 100 1000 10000
Pressure p in Pa
Load
ing
n in
mol
/kg
Experimental data
Langmuir regression
Toth regression
Adsorption of dichloromethane on MS5A at 319.4 K
0,01
0,1
1
10
0,1 1 10 100 1000 10000
Pressure p in Pa
Load
ing
n in
mol
/kg
Experimental data
Langmuir regression
Toth regression
116
Table 7.1: Comparison of Langmuir and Toth regression quality.
Adsorptive Adsorbent Temperature in
K
Langmuir
standard
deviation in %
Toth standard
deviation in %
Methanol MS5A, pellet 317.7 13.6 9.8
Methanol MS5A, grated
pellets
319.4 28.9 13.8
Methanol NaY, crystal 318.4 17.3 8.1
Methanol NaX, pellet 319.4 10.1 7.9
Dichloromethane MS5A, grated
pellets
319.4 5.8 3.8
Dichloromethane NaY, crystal 318.4 17.1 8.1
Dichloromethane NaX, pellet 316.9 3.1 2.3
7.1.3. BET surface and micropore volume
The specific surface of the solids has been measured by adsorption of nitrogen at 77
K. Figure 7.5 shows the BET isotherm for zeolite NaY in the linear form of equation
(3.7) with the relative pressure interval from 0.05 to 0.30. The intercept point of the
linear fit to the experimental data has the value b = - 0.00297 kg/mol and the gradient
m is equal to 0.152 kg/mol. With equation (3.5) the monolayer loading of the BET
isotherm nmon is equal to 6.72 mol/kg. The BET constant can be calculated by
equation (3.6) and it is c = -50.05. With equation (3.7) and the occupied area of one
nitrogen molecule amol(N2) = 0.162*10-18 m2 at 77.4 K it is SBET = 655400 m2/kg for
NaY zeolite. The other solids have been measured with the same method.
The micropore volume of the adsorbents were determined by the t-plot method. The
data for the evaluation of the micropore volume was taken from the same interval as
for the BET method, e.g. in the relative pressure interval 0.05 to 0.30. For NaY the
micropore volume has been calculated to vmicro = 0.272*10-3 m3/kg.
117
Figure 7.5: Linear BET plot for NaY zeolite.
7.2. Prediction of isotherms
7.2.1. Henry coefficient predictions
Henry coefficients were calculated with the new method (equations (5.15) and (5.65))
explained in Chapter 5 and compared with data from the literature and with own
experimental data, respectively. The model quality is exmeplarly discussed on zeolites
MS5A, MS4A, NaY and NaX.
As mentioned in Chapter 5 the potential factors for charge interactions were calculated
based on theoretical assumptions and experimental results. For induced dipole
interactions (indind, dipind and quadind) where no charge is involved on the solid side
it is assumed that
1=== quadinddipindindind fff (7.1a)
For charge involved interactions (indcha, dipcha, quadcha) it is assumed in first
approximation that
Linear BET plot for NaY zeolite
y = 0,1518377x - 0,0029745R2 = 0,9928049
0
0,01
0,02
0,03
0,04
0,05
0 0,1 0,2 0,3Relative pressure x
x/(n
*(1-
x)) i
n m
ol/k
g
118
jechjquadchajdipchajindcha Si
Alffff
⋅=== arg (7.1b)
where jSi
Al
is the aluminium to silicon ratio of adsorbent j and fcharge the potential
factor for charge involved interactions. jSi
Al
is different for every zeolite. The
parameter fcharge is equal to 1.35 for every microporous adsorbent. For MS5A it is
jSiAl
= 1, therefore findcha is assumed to be 1.35. It will be shown next that this first
approximation is very sufficient for charge interactions. In order to test the validity and
the accuracy of the new model the Henry coefficients were calculated for several
adsorption cases of gases on zeolites and compared with experimental values. With
this method it was possible to determine the unknown potential factors fi,,j in equation
(5.65).
• According to the theoretical and experimental results of Maurer [26] the induced
dipole � induced dipole potential for the zeolites also were calculated with a
potential factor findind = 1.
• Methane has no dipole moment (µi = 0), no quadrupole moment (Qi = 0) and a
polarisability of iα = 3.0*10-40 (C2*m2)/J. Adsorption on MS5A (SBET = 600000
m2/kg) at 293 K was examined regarding to [35]. The experimental Henry
coefficient is tabulated [35] as 8.8*10-6 mol/(kg*Pa). With findind = 1 in equation
(5.62) the calculated Henry coefficient is 7.93*10-6 mol/(kg*Pa) if the parameter
findcha is calculated with equation (7.1b) to 1.35. It will be shown later that the
factor findcha calculated by equation (7.1b) is adequate with respect to isotherms
investigated in this study.
• Ethene has no dipole moment (µi = 0), a quadrupole moment Qi = 1.44*10-39
C*m2 and a polarisability iα = 3.0*10-40 (C2*m2)/J. It adsorbs on NaY (SBET =
500000 m2/kg) at 305 K with a Henry coefficient of 2.56*10-3 mol/(kg*Pa) [117].
NaY has an aluminium to silicon ratio of NaYSi
Al
= 0.41. With known potential
factors findind = 1 and findcha = 1.35*0.41 = 0.554 and an assumed potential factor
fquadind = 1 the calculated Henry coefficient is 7.0*10-4 mol/(kg*Pa).
119
• Carbon dioxide has a small dipole moment µi = 2.34*10-30 Cm, a strong
quadrupole moment Qi = 1.5*10-39 C*m2 and a polarisability iα = 2.9*10-40
(C2*m2)/J. It adsorbs on MS4A (SBET = 350000 m2/kg) at 298 K with a Henry
coefficient of 1.36*10-3 mol/(kg*Pa) [22]. In adsorptive molecules with strong
quadrupole moments two dipoles in counterdirection can exist [132].
Furthermore a strong quadrupole moment can induce a dipole moment µj in the
adsorbent molecule (see chapter 5 in this thesis). The order of magnitude of the
ratio ii
iQσµ
of many polar adsorptive molecules is ≈ 1 or ijiiii Q== ,σµσµ . With
respect to this simple relationship it may be reasonable to introduce the factor
2*(2)2 = 8 in equation (5.57) in order to take into account the phenomena
described above:
( )( ) ( )33
,222
0
,222
111
2408
xkT
esQkT
x
jiadsr
atjcationsi
ads
quadcha
+−=
Φσεπε
ρ (7.1c)
An introduction of this factor 8 leads to the result that the charge potential factor
fcharge seems to have the value 1.35 for all zeolites. With an aluminium to silicon
ratio of 1 for MS4A it follows fquadcha = 1.35 and a calculated Henry coefficient of
3.4*10-3 mol/(kg*Pa).♠
• Ammonia has a strong dipole moment µi = 5.0*10-30 Cm, a strong quadrupole
moment Qi = 8.7*10-40 C*m2 and a relative small polarisability iα = 3.1*10-40
(C2*m2)/J. It adsorbs on NaY (SBET = 500000 m2/kg) at 373 K with a Henry
coefficient of 2.82*10-4 mol/(kg*Pa) [118]. With a potential factor fcharge = 1.35
and an aluminium to silicon ratio of 0.41 the Henry coefficient can be calculated
to 2.14*10-4 mol/(kg*Pa).
• Methanol has a big dipole moment µi = 5.7*10-30 Cm, a small quadrupole
moment Qi = 4.3*10-40 C*m2 and a small polarisability iα = 3.7*10-40 (C2*m2)/J. It
adsorbs on the zeolite NaY (SBET = 655400 m2/kg) at 318 K with a Henry
coefficient of 6.71*10-1 mol/(kg*Pa). The predicted Henry coefficient is 2.5*10-1
mol/(kg*Pa) if a potential factor fdipind = 1 is assumed.
♠ Another way for deriving the factor 8 is explained in the appendix C.
120
• The potential factors as shown above were taken to calculate the adsorption of
arbitrary gases on any microporous solid. The results are explained in the
following.
Figure 7.6 and 7.7 show Henry coefficients of zeolite MS4A with the adsorptives
ammonia, carbon dioxide, carbon monoxide, methane, propane, hexane, helium,
nitrogen, oxygen and benzene on zeolite MS4A. In figure 7.6 the theoretical Henry
coefficients were calculated on basis of the induced dipole- induced dipole potential.
Figure 7.6: Henry coefficients for adsorption of gases on MS4A zeolite on basis of
only induced dipole � induced dipole interactions (table 7.2).
The next figure 7.7 is based on all potentials taken into account as described in
Chapter 5 and on the potential factors. Besides benzene and hexane the prediction
works very well as can be seen in table 7.2. The van-der-Waals diameters of benzene
and n-hexane are 4.56 E-10 m and 5.18 E-10 m, respectively. It seems that these
10-7 10-6 10-5 10-4 10-3 10-2 10-1 10010-7
10-6
10-5
10-4
10-3
10-2
10-1
100
a
bcd
efgh
ij
k
l
m
no
pqrstu
Henry coefficients for MS4A zeolite
Cal
cula
ted
Hen
ry c
oeffi
cien
t in
mol
/(kg*
Pa)
Experimental Henry coefficient in mol/(kg*Pa)
23 K < Tads < 423 K
121
molecules are too big for the effective micropore opening of MS4A. The model predicts
Henry coefficient values between 3 and 5 magnitudes larger than the experimental
ones. Although sterical effects like these cannot be represented the new model works
very well over 4 orders of magnitudes in the case of the MS4A adsorption examples. It
is interesting to mention that in the case of NH3 for instance the loading is by a factor
103 higher than calculated with induced dipole � induced dipole interactions only.
The tested adsorptives cover different types of molecules with different polar
properties. Noble gases like helium have no dipole moment, no quadrupole moment
and small polarisabilities. Alkanes like methane, propane and hexane have no dipole
moment, small quadrupole moments and intermediate polarisabilities. Carbon
monoxide and carbon dioxide have small and medium dipole moments, respectively,
large quadrupole moments and small polarisabilities. NH3 has a relative big dipole
moment of 1.5 debye, a big quadrupole moment of 8.67*10-40 C*m2 but only a small
10-7 10-6 10-5 10-4 10-3 10-2 10-1 10010-7
10-6
10-5
10-4
10-3
10-2
10-1
100
abc
d
ef
ghi
j
k
l
m
no
pqr s
tu
Henry coefficients for MS4A zeolite
Cal
cula
ted
Hen
ry c
oeffi
cien
t in
mol
/(kg*
Pa)
Experimental Henry coefficient in mol/(kg*Pa)
23 K < Tads < 423 K
122
polarisability of 3.13*10-40 C2*m2/J.
Figure 7.7: Henry coefficients for adsorption of gases on MS4A zeolite (table 7.2).
Table 7.2: Comparison of predicted and experimental Henry coefficients of different
gases on MS4A zeolite (figure 7.7).
Gas Sym-
bol
Exp. Henry
coefficient in
mol/(kg*Pa)
Calc. Henry
coefficient in
mol/(kg*Pa)
Adsorption
temperature
in K
SBET in
m2/kg
Deviation in %
−
.exp
..exp
HeHeHe calc
NH3 a 1.38E-03 1.89E-04 393 350000 -38CO2 b 1.29E-03 8.23E-04 333 650000 36CO2 c 8.87E-04 7.48E-04 323 350000 16CO2 d 1.36E-03 3.39E-03 298 350000 -145CO e 1.05E-04 5.25E-05 243 650000 50CO f 3.78E-05 2.32E-05 263 650000 39CO g 1.56E-05 1.18E-05 283 650000 24CO h 7.19E-06 6.71E-06 303 650000 7He i 4.00E-06 7.35E-06 50 650000 -84He j 1.50E-05 2.32E-05 40 650000 -55He k 9.50E-05 1.62E-04 30 650000 -70He l 1.00E-03 1.81E-03 23 650000 -81C3H8 m 4.22E-05 7.16E-05 423 650000 -70N2 n 5.10E-06 3.73E-06 278 650000 27O2 o 1.40E-06 2.78E-06 278 650000 -98CH4 p 2.35E-05 2.32E-05 253 650000 1CH4 q 1.30E-05 1.30E-05 273 650000 -0.3CH4 r 9.79E-06 10.0E-06 283 650000 -2.6N2 s 1.50E-05 6.91E-06 253 650000 54N2 t 7.04E-06 4.19E-06 273 650000 40N2 u 5.30E-06 3.35E-06 283 650000 37
123
As second example for energetic heterogeneous adsorbents figures 7.8 and 7.9
show the comparison of experimental and calculated Henry coefficients for the zeolite
NaX. Table 7.3 lists the deviations in percent. Besides ammonia, methanol is a
molecule with a high dipole moment of 1.7 debye. The deviation of the calculated
Henry coefficient from the experimental one is only 41 %, whereas its deviation is
290432 % when calculated on the basis of the induced dipole � induced dipole
interaction only. A second example for polar molecules is naphthalene C10H8. It has a
high quadrupole moment of 4.5*10-39 Cm2 and a high polarisability of 1.94*10-39
C2*m2/J but no dipole moment. The deviation in figure 7.9 is 56 % whereas in figure
7.8 it is 11579 %. These data prove the efficiency of the new model very clearly. It is
interesting to note that the largest deviations are observed in the case of adsorptives
with strong quadrupole moments (NH3, CO, CO2, C2H4).
124
10-7 10-6 10-5 10-4 10-3 10-2 10-1 100 10110-7
10-6
10-5
10-4
10-3
10-2
10-1
100
101
a
b
cd
e
f
g
hij
klm
n
o
p
qr
s
tuvw
x
yz
aa
Henry coefficients for NaX zeoliteC
alcu
late
d H
enry
coe
ffici
ent i
n m
ol/(k
g*Pa
)
Experimental Henry coefficient in mol/(kg*Pa)
293 K < Tads < 623 K
Figure 7.8: Henry coefficients for adsorption of gases on NaX zeolite on basis of
induced dipole � induced dipole interactions only (table 7.3).
Table 7.3: Comparison of predicted and experimental Henry coefficients of different
gases and vapours on NaX zeolite (figure 7.9).
Gas Sym-
bol
Exp. Henry
coefficient in
mol/(kg*Pa)
Calc. Henry
coefficient in
mol/(kg*Pa)
Adsorption
temperature
in K
SBET in
m2/kg
Deviation in %
−
.exp
..exp
HeHeHe calc
C2H6 a 1.55E-04 3.10E-04 293 560000 -100
C3H8 b 2.37E-03 7.42E-03 293 560000 -213
C2H4 c 1.33E-03 3.10E-04 305,6 560000 76
C2H6 d 1.01E-04 2.09E-04 303,6 560000 -105
CH3OH e 2.25E+00 1.33E+00 323 591000 41
NH3 f 5.57E-02 7.27E-02 298,2 462000 -31
125
C10H8 g 1.56E+00 0.69E+00 523 560000 56
C10H8 h 1.50E-01 9.80E-02 573 560000 35
C10H8 i 6.67E-02 3.59E-02 603 560000 46
C10H8 j 4.10E-02 1.96E-02 623 560000 52
CO2 k 3.70E-03 6.96E-04 298,2 525000 81
CO2 l 8.49E-04 2.15E-04 323,2 525000 75
CO2 m 5.23E-04 1.53E-04 333 560000 70
C2H6 n 2.16E-04 6.65E-04 273,2 525000 -207
C2H6 o 3.17E-05 1.01E-04 323,2 525000 -217
C2H4 p 2.17E-03 5.15E-03 298,2 525000 -134
C2H4 q 6.31E-04 1.46E-04 323,2 525000 77
C2H4 r 9.30E-05 3.07E-05 373,2 525000 67
CO2 s 1.98E-03 5.37E-04 304,6 560000 73
CH4 t 6.78E-06 8.01E-06 304,5 560000 -18
N2 u 3.80E-06 1.88E-06 305,7 560000 51
O2 v 1.05E-06 1.60E-06 306,5 560000 -52
Ar w 1.27E-06 1.55E-06 304,2 560000 -22C2H6 x 9.46E-05 1.94E-04 305,6 560000 -105i-C4H10 y 3.45E-01 8.86E-02 298,2 560000 74i-C4H10 z 1.02E-02 2.28E-02 323,2 560000 -124C3H6 aa 7.88E-03 5.29E-03 303 560000 33
126
10-7 10-6 10-5 10-4 10-3 10-2 10-1 100 10110-7
10-6
10-5
10-4
10-3
10-2
10-1
100
101
a
b
cd
e
f
g
hi
j
klm
n
o
p
q
r
s
t
uvw
x
y
z
aa
Henry coefficients for NaX zeoliteC
alcu
late
d H
enry
coe
ffici
ent i
n m
ol/(k
g*Pa
)
Experimental Henry coefficient in mol/(kg*Pa)
293 K < Tads < 623 K
Figure 7.9: Henry coefficients for adsorption of gases on NaX zeolite (table 7.3).
7.2.2. Isotherm predictions
In Chapter 6 a new model was introduced to calculate the whole adsorption isotherm
from the linear region until saturation. In this Section the model will be verified based
on literature and experimental data. Henry coefficients of the model are calculated as
shown in Chapter 6 and verified in Section 7.2.1. The saturation data which is the
second main boundary condition is tabulated in tables in addition to the graphic
presentation.
In figure 7.10 the prediction of isotherms for active carbon coal is shown. Active
carbon coal is an energetic homogeneous adsorbent. In this case the adsorption
attraction of arbitrary gases is dominated by the reduced Hamaker energy.
127
Active carbon coal can be modelled very well with the prediction theory of Chapter 6.
It it assumed that the parameter C is a constant with C = 0.55*10-3. This is verified
also in Section 7.3.3. The calculated Henry coefficients of various types of molecules
differ at most 19 % (in the case of carbon dioxide) from the experimental data. Also
the saturation region is modelled very well. It seems that the new model can cover
any arbitrary gas if adsorbed on energetic homogenous adsorbents like active
carbon.
Figure 7.10: Experimental and predicted adsorption isotherms of active carbon coal.
Experimental data are for nitrogen (303.15 K, −), methane (303.15 K, ▲), carbon
dioxide (303.15 K, ♦ ), propane (298 K, ■) and methanol (298.15 K, • ), −, --
calculated.
Adsorption of arbitrary gases on active carbon
1E-6
1E-5
1E-4
1E-3
1E-2
1E-1
1E+0
1E+1
1E+2
1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8
Pressure p in Pa
Load
ing
n in
mol
/kg
CH3OH
CO2
N2
CH4
C3H8
298 K < Tads < 303K
128
Table 7.4: Properties of adsorptives and adsorbents in figure 7.10.
Adsorptive
and cited
literature
Heexp in
mol/(kg*Pa)
Hecalc in
mol/(kg*Pa)
Tads
in K
SBET in
m2/kg
vmicro in
m3/kg micro
jBET
vS σ
( )KN iA
300
3
βσ⋅
Nitrogen
[37]
4.12E-6 4.25E-6 303 1105000 4.0E-4
[120]
0.967 0.28
Methane
[37]
1.33E-5 1.32E-5 303 1105000 4.0E-4
[120]
0.967 0.40
Carbon
dioxide
[37]
6.78E-5 5.47E-5 303 1105000 4.0E-4
[120]
0.967 0.48
Propane
[124]
3.35E-2 6.99E-3 298 1200000 4.5E-4
[125]
0.933 0.52
Methanol
[119]
8.04E-3 8.35E-3 298 1440000 6.5E-4
[121]
0.775 0.85
In order to test the validity and the accuracy of the new model adsorption isotherms
over the entire range from the Henry region up to saturation have been calculated for
special adsorptives on zeolites.
Figure 7.11: Adsorption of n-hexane on silicalite-1 at 298 K [126]. −, -- calculated.
α i-dominant adsorption
1E-2
1E-1
1E+0
1E+1
1E+2
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4
Pressure p in Pa
Load
ing
n in
mol
/kg
Tads = 298 K
129
Figure 7.11 shows an αi-dominant adsorption case of n-hexane on silicalite-1 [126].
n-hexane possesses a high polarisability but zero polarities. Silicalite-1 has no
cations in its structure, therefore it has no charge involved interactions, The example
shows that the αi-dominant adsorption case is covered very well by the model.
An additional approach for Hamaker-dominant systems will be presented here. In
figure 7.12 the dimensionless loading n/(He*p) is plotted versus a dimensionless term
which contains the loading n, the saturation loading n∞, the critical data pc and Tc of
the adsorptive and the Hamaker energy Haj. The diagram is valid for Hamaker-
dominant systems (for instance activated carbon as adsorbent). The plot is similar to
the diagram published in [27] as figure 3.22. However, in the figure of [27] the
expression of (He*p)/n is plotted versus an abscissa expression with the number of
molecule layers nSN
NBET
iA2σ
= . In figure 7.12 the term
−
∞nn1 can be interpreted as
the fraction of the micropore volume not occupied by the adsorbate yet. Furthermore,
the expression is powered by the exponent 2/3 in order to obtain the Hamaker
energy Haj with the exponent one instead of 3/2 [27]. The straight line in figure 7.12 is
supported by experimental results of N2, CH4, C3H8, C3H6, C6H14 and ethylacetate
adsorbed on activated carbon in the temperature range between 25°C and 93°C and
delivers the relationship
−
⋅−−=
∞
−
32
23
34 1ln1036.4ln
nn
TT
pHa
pHen
ads
C
Cj
j
σ (7.1)
As can be seen the adsorption isotherm for Tads depends on
• the Henry coefficient He,
• the saturation loading n∞ and on
• the physical properties of the adsorptive and the adsorbent.
130
Figure 7.12: The dimensionless group pHen is plotted against the dimensionless
group 3
22
3
3 1ln
−
−
∞nn
TT
pHa
ads
C
Cj
j
σ.
According to Mersmann et.al. [27] the Henry coefficient He can be approximated by
ℜ≈
21
3, 4.0exp
Cj
j
ads
C
ads
BETji
pHa
TT
TS
Heσ
σ (7.2)
1E-05
1E-04
1E-03
1E-02
1E-01
1E+00
0 500 1000 1500 2000
Dim
ensi
onle
ss lo
adin
g n/
(He
p).
32
23
3 1ln
−
−
∞nn
TT
pHa
ads
C
Cj
j
σ
Lower limit
Upper limit
131
In this equation the diameter σi,j and the BET surface are additional parameters. A
combination of the two equations (7.1) and (7.2) results in
−
⋅−−
ℜ≈
∞
−
32
23
34
21
3
,
1ln1036.4exp
4.0exp
nn
TT
pHa
pHa
TT
TS
pn
ads
C
Cj
j
Cj
j
ads
C
ads
BETji
σ
σσ (7.3)
In figure 7.12 most of the experimental data are located in between the upper and
lower line. With respect to the accuracy of an adsorption isotherm calculated by
equation (7.3) the Hamaker energy Haj plays the biggest role. In order to obtain an
accuracy of +- 20 % with respect to the entire isotherm an accuracy of approximate
+- 5% of the volumetric Hamaker energy
3j
jHaσ
is necessary. This may explain
some deviations between experimental and predicted data.
Figure 7.13 shows the adsorption of other nonpolar adsorptives on zeolites NaX, NaY
and MS5A. Table 7.5 lists the properties of the adsorptives and adsorbents in figure
7.13. The new model is very successful also in predicting adsorption isotherms in the
case of nonpolar gases and energetic heterogeneous adsorbents. It has to be
mentioned that the success of the model depends strongly on reliable data of the
Hamaker energy, the BET surface and the micropore volume of the adsorbent. In the
case of zeolites the micropore volume was taken from Kast [24]. The BET surfaces
for NaX and NaY in figure 7.13 are average values of literature data. For practical
purposes the predicted isotherms are calculated down to the equilibrium pressure 1
Pa. Thus, it is possible to read the Henry coefficient directly from the intercept of the
predicted isotherm with the y-axes. For all adsorbents, zeolites and active carbon, the
saturation loading seems to be achieved at 10+8 Pa.
132
Table 7.5: Properties of adsorptives and adsorbents in figure 7.13.
Adsorptive and
cited literature
Heexp in
mol/(kg*Pa)
Hecalc in
mol/(kg*Pa)
Tads
in K
SBET in
m2/kg
vmicro in
m3/kg micro
jBET
vS σ
( )KN iA
300
3
βσ⋅
Methane/NaY
[115]
3.75E-6 4.42E-6 298 500000 3.0E-4
[24]
0.70 0.40
Methane/NaX
[7]
6.78E-6 8.01E-6 304 560000 3.0E-4
[24]
0.72 0.40
Ethane/MS5A
[36]
1.18E-4 1.09E-4 313 727000 2.5E-4
[24]
1.1 0.48
Propane/MS5A
[36]
1.95E-3 1.40E-3 313 727000 2.5E-4
[24]
1.1 0.52
Figure 7.13: Experimental and predicted adsorption isotherms of zeolites.
Experimental data are for methane/NaY (298 K, • ), methane/NaX (304.5 K, ■),
ethane/MS5A (313 K, ♦ ) and propane/MS5A (313 K, ▲). −, -- calculated.
Adsorption of nonpolar gases on zeolites
1E-6
1E-5
1E-4
1E-3
1E-2
1E-1
1E+0
1E+1
1E+2
1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8
Pressure p in Pa
Load
ing
n in
mol
/kg
298 K < Tads < 313 K
propane/MS5A
ethane/MS5A
methane/NaX
methane/NaY
133
Methanol and ammonia are adsorptives with high dipole moments but rather small
polarisabilities. Furthermore the quadrupol moment of ammonia is very high.
Especially methanol is an adsorptive which offers the possibility to test the validity of
the equations (3) and (4) of figure 5.5. Figure 7.14 shows the adsorption of methanol
and ammonia [18] on the zeolites NaY and NaX, respectively.
Figure 7.14: Adsorption of methanol on NaY (318 K, ■) and ammonia on NaX (298
K, ♦ ). −, -- calculated.
The adsorptives ethene and carbon dioxide are characterised by large quadrupole
moments but zero (ethene) or moderate (carbon dioxide) dipole moments and rather
moderate polarisabilites. Calculated adsorption isotherms of these adsorptives can
be used for testing the validity and accuracy of equations (5) and (6) in figure 5.5.
Figure 7.15 shows the adsorption of ethene on MS5A [27] at 303 K and the
adsorption of carbon dioxide on NaY [117] at 305 K.
µi-dominant adsorption
1E-4
1E-3
1E-2
1E-1
1E+0
1E+1
1E+2
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5
Pressure p in Pa
Load
ing
n in
mol
/kg
298 K < Tads < 318 K
NH3/NaX
methanol/NaY
134
Figure 7.15: Adsorption of ethene on MS5A [27] at 303 K and the adsorption of
carbon dioxide on NaY [117] at 305 K. −, -- calculated.
As overall case figure 7.16 and table 7.6 show the adsorption of any polar gases on
energetic heterogeneous adsorbents.
Qi-dominant adsorption
1E-2
1E-1
1E+0
1E+1
1E+2
1E+3 1E+4 1E+5 1E+6 1E+7
Pressure p in Pa
Load
ing
n in
mol
/kg
303 K < Tads < 305 K
CO2/NaY
C2H4/MS5A
135
Figure 7.16: Experimental and predicted adsorption isotherms of zeolites.
Experimental data are for CO2/MS5A (343 K, ■), NH3/NaX (298 K, • ), p-xylene/NaY
(523K, 7) and CH2Cl2/MS5A (319.35 K, ♦ ). −, -- calculated.
Table 7.6: Properties of adsorptives and adsorbents in figure 7.16.
Adsorptive
and cited
literature
Heexp in
mol/(kg*Pa)
Hecalc in
mol/(kg*Pa)
Tads in
K
SBET in
m2/kg
vmicro in
m3/kg micro
jBET
vS σ
( )KN iA
300
3
βσ⋅
CO2/MS5A
[36]
3.93E-4 2.17E-4 343.0 727000 2.5E-4
[24]
1.1 0.48
NH3/NaX
[18]
5.57E-2 7.27E-3 298 462000 3.0E-4
[24]
0.59 0.6
CH2Cl2/MS5A
[own]
6.14E-1 1.06E-0 319.4 464000 2.3E-4 0.77 0.58
p-xylene/NaY 8.93E-3 2.78E-3 523 500000 see text
Adsorption of polar gases on zeolites
1E-4
1E-3
1E-2
1E-1
1E+0
1E+1
1E+2
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6
Pressure p in Pa
Load
ing
n in
mol
/kg CH2Cl2/MS5A
CO2/MS5A
298 K < Tads < 523 K
NH3/NaX p-xylene/NaY
136
The isotherms of figure 7.16 show a highly polar adsorptive (dichlormethane) with a
dipole moment of 1.6 debye and three molecules with high polarisabilities and high
quadrupole moments (NH3, p-xylene and CO2). In the case of p-xylene the saturation
loading was calculated with the adsorbate density as given in the literature, 1.08*10+3
kg/m3 [116] and with equation 6.4.
All the predicted isotherms fit the experimental data very well in the Henry region and
the saturation region. By the different systems it is verified that the new model can
predict the adsorption of any kind of gas, polar or unpolar, on any kind of
microporous adsorbent, energetic homogeneous or heterogeneous. However, in the
figure 7.14 the system methanol/NaY show higher loadings than calculated in the
transition range.
In the next Sections the parameters of the new model will be verified.
7.3. Model verifications
7.3.1. Refractive index verification
The refractive index is an important parameter for calculating the Hamaker constant
of microporous solids like zeolites. In this work the Clausius-Mosotti relation was
used to calculate this parameter. In Chapter 3 additional ways were introduced for
calculating the refractive index that should be reviewed in comparison to the used
method.
As can be seen in equation (3.34) and also as explained in Chapter 3 the Clausius-
Mosotti formula describes a border case of the general formula of Marler. In this case
the electrical fields of the solid ions do not overlap each other. The other case is
described by the formula of Newton-Drude. Here, the overlapping field compensates
the Lorentz field exactly. All other cases are between these two border cases, which
can be described by the parameter b. So the refractive index can be rewritten as a
function of b and the solid parameters.
By introducing a dimensionless solid parameter P* with
137
solid
solidm
solid
solidAsolid
MR
MN
Pρ
ερα 3
0
* == (7.4)
the refractive index is
π
π
41
41
*
**
Pb
PbPn
−
−+= (7.5)
The two border cases are for Marler with b = 0
*1 PnMarler += (7.6)
and for Clausius-Mosotti with 3
4π=b (7.7)
31
321
*
*
P
P
n−
+= (7.8)
For MS5A the solid parameter P* is equal to 9,83*10-1 with the assumptions made in
Chapter 3 and in table 7.7.
138
Table 7.7: Solid parameters of zeolite MS5A.
Parameter Value
Chemical Structure Ca5Na2((AlO2)12(SiO2)12) Molar refraction Rm in m3/mol regarding to
chemical structure 4.79*10-4 Apparent density ρsolid in kg/ m3 1150
Molar mass Msolid in kg/mol regarding to
chemical structure
1.68
Parameter P* 0.983
The function of n in dependence of parameter b (equation 7.5) with the two border
cases is plotted in figure 7.17 for zeolite MS5A.
Refraction index of MS5A in dependence from the electronic overlap parameter b
1,2
1,4
1,6
1,8
0 1 2 3 4
Electronic overlap parameter b (-)
Ref
ract
ion
inde
x n
of
MS
5A (-
)
Figure 7.17: Refractive index of MS5A in dependence on the electronic overlap
parameter b.
139
The two border cases result in a refractive index nmin = 1,41 and nmax = 1,57. Following
the equations in Chapter 3 the dependence of the Hamaker constant Ha on the
refractive index n is described by
21~ 2
2
+−
nn
solidα and α
ν 1~0 and 20~~ ανβHa (7.9a)
Therefore, it follows that
5,1
2
2
21~
+−
nnHa (7.9b)
The two border cases lead to a difference of ca. 35 % of the numerical value of the
Hamaker constant of MS5A. The maximum value in this work, HaMS5A,max = 7.66*10-20
J, is calculated with Clausius-Mosotti whereas the minimum value is HaMS5A,min =
5.0*10-20 J.
The Henry coefficient depends on the Hamaker constant. For further calculations
propane adsorption on MS5A is taken as example. The parameters for this case are Tc
= 369.8 K, Tads = 343 K, pc = 4.25*106 Pa and σMS5A = 3.8*10-10 m. It follows with
cjads
c
pHa
TT
IP 3maxmin/
maxmin/2
πσ= (7.10)
and
ILHe ~ (7.11)
the percentage difference of the Henry coefficient with
( ) ( )( ) %100
expexpexp
%100(%)2
22
min1
min1max1
min
minmax ⋅−
=⋅−
=∆ p
pp
IPpIPpIPp
ILILIL
He (7.12)
140
For the given example the percentage difference is ca. 550 %.
The discussion in the previous section has shown that the theoretical and experimental
Henry coefficients differ much less. So it seems that the Clausius-Mosotti formula is
the right choice for calculating the refractive index and also the Hamaker constant.
Another reason is that the Clausius-Mosotti formula is described as standard equation
in widely accepted literature [6].
7.3.2. Dependency on the saturation degree
In Chapter 6 the parameter C in equation (6.3) was assumed to be constant in a first
approximation. In this Section a linear dependency of C on the saturation degree nS* is
discussed with
bnnaC +=∞
(7.13)
Approximating C between the border values 0.6*10-3 and 0.5*10-3 for saturation
degrees 0 and 1, respectively, the parameters of equation (7.13) are equal to a = -
0.1*10-3, b = 0.6*10-3. The differential equation (6.3) turns then to
−
++
==
∞
∞
∞ nnnn
pHa
TT
bnna
np
dpdn
pdnd
cj
j
ads
c
11
1)(ln)(ln
23
3σ
(7.14)
Equation (7.14) can be rewritten with the term B� defined by
23
3
=′
cj
j
ads
c
pHa
TT
Bσ
(7.15)
into
141
)(ln1
1)(ln pdbnna
nnnn
Bnd =
+
−′+
∞∞
∞ or (7.16)
Hepndnb
nna
nnnn
Bn lnln1
ln +=
+
−′+ ∫
∞∞
∞ (7.17)
This can be reduced to
dnnnnb
nna
BnpHe
∫
−
+′=
∞
∞∞
1ln
2
(7.18)
Integration of the right hand side of equation (7.18) leads to
−−+
−−−′=
∞∞∞ nnb
nn
nnaB
nHep 1ln1lnln (7.19)
or
( )∞∞
′−
−+′−=
nnaB
nnbaB
nHep 1lnln (7.20)
or
( )
∞
+′
∞
′−=
−
nnaB
nn
nHep
baB
1ln (7.21)
It is therefore
142
( )
′−
−
=∞
+′
∞
nnaB
nnHe
np baB exp
1
1 (7.22)
or in dependence on the saturation degree ∞
=nnns
* it follows that
( ) ( ) ( )*
*exp
11
SbaBS
anBnHe
np ′−−
= +′ (7.23)
This modified model is verified on the adsorption example carbon dioxide on active
carbon, the same example as in figure 7.10. The result is shown in figure 7.18. As
can be seen in figure 7.18 there is no significant difference between the two cases.
So it is assumed that the parameter C in equation (6.3) is a constant being
independent on the saturation degree.
143
Figure 7.18: Comparison of modified model with the original prediction isotherm.
7.3.3. The London potential
In Chapter 5 the assumption was made that the induced dipole � induced dipole
interaction energy can be derived from the London potential also. In [24] the London
potential for two dissimilar atoms is calculated by
( )( ) ( )ji
jijiL
h
rrE
νννν
πεαα
+−=
62042
3 (7.24)
where iα is the polarisability of the gas molecule in (C2*m2)/J and jα the polarisability
of the solid molecule. It can be replaced by solidα , the mean average polarisability of
the solid continuum molecule calculated by equation (3.97). νj is assumed to be the
Adsorption of CO2 on active carbon
1E-6
1E-5
1E-4
1E-3
1E-2
1E-1
1E+0
1E+1
1E+2
1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6 1E+7 1E+8
Pressure p in Pa
Load
ing
n in
mol
/kg
CO2, 303,15 K, exp.
CO2, 303,15 K, pred.
CO2, 303,15 K, Henry regionpred.CO2, 303,15 K, Saturation regionpred.Modified Model, pred.
144
frequency of the solid electrons in the ground state in 1/s and can be calculated by
(3.98). Furthermore it is defined that
ii hI ν= (7.25)
where Ι i is the ionisation energy of the adsorptive molecule in J. It is tabulated in
literature [6]. Inserted in equation (7.24) it is
( )( )
+
−=
0
062
0423
ν
νπε
αα
hII
rrE
i
isolidiL (7.26)
With the general interaction energy as calculated in equation (5.40) the induced dipole
� induced dipole potential is
( ) ( )∫∞
+
=Φijz
atjLindind drrrEzσ
πρ 2, 4 (7.27)
or
( )( )33
,0
20
0, 1
11
8 xhI
kT
IkT
x
jiiads
isolidiatj
ads
indind
+
+⋅
−=Φ
σνπε
νααρ (7.28)
It can be defined an alternative interaction parameter for the induced dipole � induced
dipole interaction:
3,
020
0,
1
8 jiiads
isolidiatjindind
hI
kT
IA
σνπε
νααρ
+⋅
−= (7.29)
In table 7.8 the values for Aindind calculated by equations (5.62) and (7.29) are
compared for the case of adsorption of gases on zeolite MS5A.
145
Table 7.8: Interaction parameter Aindind calculated with the two equations (5.62) and
(7.29)
Molecule Ionisation
energy Ii in
10-18J
Tads in K Aindind with
equation
(5.62)
Aindind with
equation
(7.29)
N2 2.5 303 3.8 3.7
CO 2.25 303 5.2 4.0
NO 1.48 383 4.1 2.7
CH4 2.02 343 5.8 4.5
O2 1.93 293 5.3 3.5
Xe 1.94 195 13.7 10.8
Kr 2.24 195 10.2 7.9
C2H4 1.68 303 9.3 6.5
CHF3 2.22 303 10.0 5.8
NH3 1.62 298 9.0 5.4
The values are very similar. The assumption in Chapter 5 is proved now. In this work
equation (5.62) was taken to be consistent with the literature [26, 27].
7.3.4. Potential energy comparison and formula overview
The introduced model according to equations (5.61) and (6.3) is a theoretical basis
for predicting isotherms of gases on microporous solids. It is derived from the
potential energies which are attractive between the adsorptive and the adsorbent. In
this section the main formulae of the different cases will be tabulated for engineering
purposes.
As example for the acting different attraction energies the adsorption between
methanol and NaY zeolite will be examined quantitatively. Between molecules of this
adsorption pair all kinds of potential energy can exist. Therefore, it is ideal for the
comparison of the adsorption forces. As discussed in Chapter 5 the Henry coefficient
of polar gases with energetic heterogeneous microporous solids is described by the
integral equation (5.61). For comparing the energies of attraction the percentage of
146
the energies is calculated when the adsorptive is attached on the solid surface, in
other words the distance z between adsorptive and adsorbent is equal to 0. Table
7.9a gives a detailed overview of the situation. With an adsorption temperature of
Tads = 318.4 K, the energy unit is k*Tads = 4.395E-21 J. The other parameters are: σj =
4.2.10-10 m, He = 0.671.mol/(kg.Pa), SBET = 655400 m2/kg, µi = 5.67.10-30 C.m, αi =
3.66.10-40 C2.m2/J, Qi = 4.34.10-40 C.m2, Tc = 512.6 K, pc = 8090000 Pa, s = 2.4, scations
= 0.09, 41.0=
NaYSiAl , 28
atj, 103.3 ⋅=ρ 1/m3, Haj = 8.4.10-20 J and α j = 2.82.10-40
C2.m2/J.
In the case of methanol adsorption on NaY at 318.4 K there are two potentials, which
contribute 97 % of the overall attraction energy, respectively the induced dipole �
induced dipole and the permanent dipole � charge interactions. From the other four
the two quadrupole interactions only make up 0.42 % of the overall potential. A
second calculation for CO2 on NaX at 304.55 K (table 7.9b) shows that the
quadrupole - charge interaction contributes approximately 12,3 % of Φ . These two
results make it clear that three potentials are eminent for the calculation of Henry
coefficients: Always the induced dipole � induced dipole interaction indind and
dependent from the adsorptive and the adsorbent the permanent dipole � charge
interaction dipcha and/or the quadrupole moment � charge interaction quadcha.
The main formulae of this work are collected in table 7.10. These are the refractive
index and Hamaker constant of zeolites, the Henry coefficient for the adsorption of
gases in the case of Hamaker dominant attraction energies and also for the case of
polar-polar adsorption pairs and finally the new model for predicting the whole
adsorption isotherm. The section in which the formula appears in a detailed
description is also tabulated.
14
7 Ta
ble
7.9a
: Attr
actio
n en
ergi
es b
etw
een
met
hano
l and
NaY
at a
dis
tanc
e z
= 0
of th
e ad
sorp
tive
from
the
adso
rben
t sur
face
.
Indu
ced
dipo
le
�
indu
ced
dipo
le
inte
ract
ion
(indi
nd)
Indu
ced
dipo
le �
char
ge in
tera
ctio
n
(indc
ha)
Perm
anen
t
dipo
le �
ind
uced
dipo
le i
nter
actio
n
(dip
ind)
Perm
anen
t dip
ole
�
char
ge
inte
ract
ion
(dip
cha)
Qua
drup
ole
mom
ent
� in
duce
d di
pole
inte
ract
ion
(qua
dind
)
Qua
drup
ole
mom
ent
� ch
arge
in
tera
ctio
n
(qua
dcha
)
()
cc
jj
j
indind
Tp
Ha
k
3 ,
243
0 σπ
⋅
⋅=
Φ
()
ji
indcha
r
cations
iindcha
f
es
,
atj,
22 0
22 1
810 σ
ρε
πε
α ⋅
⋅⋅
⋅=
Φ
()
ji
dipind
r
ji
dipind
f,
atj,
22 0
2
1
12
0 σρ
επ
εαµ
⋅
⋅⋅
=
=Φ
()
ji
dipcha
r
cations
i
ads
dipcha
f
es
kT
,
atj,
22 0
22
2
1
12
10 σ
ρε
πε
µ ⋅
⋅⋅
⋅
⋅=
Φ
()
()5
,
atj,2
2 0
2
1
403
10
ji
quadind
rij
ads
quadind
fQ
kT
σ
ρε
επα
⋅
⋅⋅
⋅=
Φ
()
()3
,
atj,2
2 0
22
2
1240
10
ji
quadcha
rications
ads
quadcha
fQ
es
kT
σ
ρε
πε
⋅
⋅⋅
⋅=
Φ
()J
indind
2010
7,6
0 −⋅
=
=Φ
()
Jindcha
2110
07.2
0−
⋅=
=Φ
()
Jdipind
2210
98.5
0−
⋅=
=Φ
()
Jdipcha
2010
76.2
0−
⋅=
=Φ
()
Jquadind
2310
00.2
0−
⋅=
=Φ
()
Jquadcha
2210
10.4
0 −⋅
=
=Φ
()ads
indind
kT⋅=
=Φ
21,15
0
()ads
indcha
kT⋅=
=Φ
47.00
()
ads
dipind
kT⋅=
=Φ
14.0
0
()ads
dipcha
kT⋅=
=Φ
27.6
0
()ads
quadind
kT⋅=
=Φ
005
.0
0
() ads
quadcha kT⋅
=
=Φ
1.0
0
()Φ⋅
=
=Φ
%5.68
�0
indind
() Φ⋅
=
=Φ
%1.2
�0
indcha
()
Φ⋅⋅
=
=Φ
%6.0
�0
dipind
()
Φ⋅=
=Φ
%3.28
�
�0
dipcha
()
Φ⋅=
=Φ
%02.0
�
�0
quadind
() Φ⋅
=
=Φ
%4.0�
�0
quadcha
14
8 Ta
ble
7.9b
: Attr
actio
n en
ergi
es b
etw
een
CO
2 and
MS5
A at
a d
ista
nce
z =
0 of
the
adso
rptiv
e fro
m th
e ad
sorb
ent s
urfa
ce.
Indu
ced
dipo
le
�
indu
ced
dipo
le
inte
ract
ion
(indi
nd)
Indu
ced
dipo
le �
char
ge in
tera
ctio
n
(indc
ha)
Perm
anen
t
dipo
le �
ind
uced
dipo
le i
nter
actio
n
(dip
ind)
Perm
anen
t dip
ole
�
char
ge
inte
ract
ion
(dip
cha)
Qua
drup
ole
mom
ent
� in
duce
d di
pole
inte
ract
ion
(qua
dind
)
Qua
drup
ole
mom
ent
� ch
arge
in
tera
ctio
n
(qua
dcha
)
()
cc
jj
j
indind
Tp
Ha
k
3 ,
243
0 σπ
⋅
⋅=
Φ
()
ji
indcha
r
cations
iindcha
f
es
,
atj,
22 0
22 1
810 σ
ρε
πε
α ⋅
⋅⋅
⋅=
Φ
()
ji
dipind
r
ji
dipind
f,
atj,
22 0
2
1
12
0 σρ
επ
εαµ
⋅
⋅⋅
=
=Φ
()
ji
dipcha
r
cations
i
ads
dipcha
f
es
kT
,
atj,
22 0
22
2
1
12
10 σ
ρε
πε
µ ⋅
⋅⋅
⋅
⋅=
Φ
()
()5
,
atj,2
2 0
2
1
403
10
ji
quadind
rij
ads
quadind
fQ
kT
σ
ρε
επα
⋅
⋅⋅
⋅=
Φ
()
()3
,
atj,2
2 0
22
2
1240
10
ji
quadcha
rications
ads
quadcha
fQ
es
kT
σ
ρε
πε
⋅
⋅⋅
⋅=
Φ
()J
indind
2010
67.4
0−
⋅=
=Φ
()
Jindcha
2110
97.1
0−
⋅=
=Φ
()
Jdipind
2210
42.1
0−
⋅=
=Φ
()
Jdipcha
2110
91.5
0−
⋅=
=Φ
()
Jquadind
2210
18.4
0 −⋅
=
=Φ
()
Jquadcha
2110
75.7
0 −⋅
=
=Φ
()ads
indind
kT⋅=
=Φ
11,11
0
()ads
indcha
kT⋅=
=Φ
47.00
()
ads
dipind
kT⋅=
=Φ
03.0
0
()ads
dipcha
kT⋅=
=Φ
41.1
0
()ads
quadind
kT⋅=
=Φ
09.0
0
() ads
quadcha kT⋅
=
=Φ
84.1
0
()Φ⋅
=
=Φ
%3.74
�0
indind
() Φ⋅
=
=Φ
%1.3
�0
indcha
()
Φ⋅⋅
=
=Φ
%2.0
�0
dipind
()
Φ⋅=
=Φ
%4.9�
�0
dipcha
() Φ⋅
=
=Φ
%7.0�
�0
quadind
()
Φ⋅=
=Φ
%3.12 �
�0
quadcha
149
Table 7.10: Main formulae and cases for the adsorption of gases on microporous solids.
Description of
formula
Formula Section
Clausius-Mosotti
relationship 21
3 0 +−
=r
rA MNεε
ρεα
3.3.5.1
Refractive index n of
zeolites
MRMR
nm
m
ρ
ρ
−
+=
1
21
3.3.5.2
Hamaker constant
Haj of solids βρπ 2
atj,2=jHa , M
Nn Aa
ρρ =atj,,
απν 1
20em
e=,
( )213
2
2
0 +−=
nn
NnM
Aa ρεα
2
00 44
3
=
επανβ hs
3.3.6.2
Henry coefficient He
for Hamaker-
dominant adsorption jiBET
ads
STHe
IL,σ
ℜ≡
, cjj
j
ads
c
pHa
TT
IP 3,
2σπ
≡
( )21exp pIPpIL = , p1 = 0.331
p2 = 1.180
5.2.1
Henry coefficient He
for adsorption of
polar adsorptives on
energetic
heterogeneous
solids
( )
( )
( )
∫
−
++
+
+++
+++
=max
0
5
3 1
1
1
1
expx
quadind
quadladdipindindind
indladdiplad
dx
x
Ax
AAAxAA
IL
xmax = 2
5.3.8
Adsorption isotherm
for gases on
microporous solids
B
nnHe
np
∞−=
11 ,
23
3
=
cj
j
ads
c
pHa
TT
CBσ
( )ads
micro
Tv
nβ
=∞ , ( ) ( )bvdWbc
badsbads vv
TTTT
vT −
−−
+=β
C = 0.55 10-3
6.1 and 6.2.3
150
8. Summary The main objective of this thesis is the theoretical prediction of adsorption equilibria
of single gaseous substances on microporous solids like active carbon or zeolites
(e.g. MS4A, MS5A, NaX and NaY). A novel equilibrium model was derived from a
differential equation considering two boundary conditions. These boundaries were
the Henry coefficient for diluted gases and the saturation loading for highly
concentrated gases.
A historical overview on adsorption is presented in Chapter 3 which also addresses
some important chemical and technical aspects. An essential prerequisite for the
predicition of adsorption equilibria is a detailed knowledge of the properties of the
solid adsorbents. The characterisation of microporous adsorbents is treated in
Section 3.3. The novel equilibrium model is compared with classical adsorption
theories as the models of Langmuir and Toth. Techniques for the measurement of
adsorption equilibria (e.g. adsorption isotherms) are explained also.
The Henry coefficient which describes the adsorption isotherm of diluted gases
depends on the properties of both the gaseous (adsorptive) and the solid (adsorbent)
phase. However, the key parameter of the intermolecular interaction forces, the
Hamaker constant, is unknown for most zeolites. In Chapter 3, an approximated
Hamaker and de Boer model for predicting the Hamaker constant of microporous
solids is developed on the basis of the refractive index.
The refractive index is determined by the Clausius-Mosotti relationship between the
relevant optical and density parameters of the solid adsorbent. For zeolitic
adsorbents, the optical parameters can be derived from the polarisability of ions in
the zeolite crystal.
Experimental methods for the determination of adsorption equilibria are presented in
Chapter 4 of this thesis. Besides classical techniques some novel ones are treated in
detail, e.g. the zero length column method.
151
An important result of the thesis is the finding that the adsorption of unpolar as well
as polar gases on energetical heterogeneous adsorbents is controlled by the
Hamaker constant. Therefore, the dominant mechanism is the induced dipole �
induced dipole interaction. For this case the dimensionless Henry curve was
numerically evaluated because the differential equation cannot be solved analytically.
The numerical results were correlated by an exponential function. The adsorption of
polar gases on energetical heterogeneous adsorbents depends on more than two
parameters. It was not possible to correlate the numerical results by simple
regression. Nevertheless, a suited algorithm is presented in Chapter 5.
In Chapter 6, the prediction model for all cases was developed by integrating the
fundamental differential equation under consideration of two boundary conditions,
e.g.. the Henry coefficient and the saturation loading.
In Chapter 7 the novel model was compared with literature and experimental data.
The Toth isotherm model was used to calculate the experimental Henry coefficients.
The calculated Henry coefficients of the novel model fit very well with experimental
data over six orders of magnitudes. As second step the predicted adsorption
equilibria were compared with experimental data from the literature. The novel model
fits the isotherms very well beginning at the Henry region and ending at the saturation
region. Next in Chapter 7, the assumptions for the calculation of the adsorption model
parameters, e.g. the refractive index, were verified. For engineering purposes it is
important to know which attraction potentials are of theoretical interest only and
which are dominant for calculating the Henry coefficient. As a result it was shown that
the induced dipole � induced dipole interaction is always eminent. For polar gases
additionally the permanent dipole � charge and/or the quadrupole moment � charge
interaction have to be taken into account. All other induced dipole interactions have
minimal impact and can be neglected.
In the appendix the set-up for own experimental data will be explained.
152
Appendix
Appendix A: Isotherms in transcendental form
Maurer [26] introduced a generalized empirical equation that allows for a priori
predictions of complete adsorption isotherms for active carbon with
( )( )
=
kgmoln
barpKT
SfnpHec
cBET 2exp (A.1)
where He is the Henry coefficient in mol/(kg*Pa), p the adsorption equilibrium
pressure in Pa, n the loading of the adsorbent in mol/kg, Tc the critical temperature of
the adsorptive in K and pc the critical pressure of the adsorptive in bar. f(SBET) is a
function of the BET surface area SBET and follows from
( )
gmS
gm
Sf
BET
BET 2
2
6167367,0 +−= (A.2)
The Henry coefficient can be determined by the following equation for energetic
homogeneous adsorbents, derived in Chapter 5 with
( )
ℜ=
ℜ=
ℜ=
2
231
,1
,, 2expexp
p
cj
j
ads
c
ads
BETjip
ads
BETji
ads
BETji
pHa
TT
pTS
IPpTS
ILTS
Heσπ
σσσ
(A.3)
where all the coefficients are in SI units and explained already in Chapter 5. The
parameters are p1 = 0.3313 and p2 = 1.180. The loading n can be calculated by given
partial pressure p.
153
Equation (A.2) cannot be solved analytically for n. The given function has the
structure
( ) ( )xuxxy exp= (A.4)
The inverse function of equation (A.4) is the so-called Lambert function W with
( )uyuWx = (A.5)
where y is a substitution for
pHey = (A.6)
and the parameter u a substitution for
( )( )2barp
KTSfu
c
cBET= (A.7)
Rechid [104] introduced a mathematical approach for describing the Lambert function
with analytically solvable functions. It is therefore
( ) ( )( ) ( )( )( )u
yuyuyun 111lnlnln3948.211lnln1037.21ln0004.1 ++++++−+= (A.8)
The adsorption isotherm for active carbon can be calculated straight forward
analytically only with knowledge of the properties of adsorptive and adsorbent. As
example the isotherm of ethane on active carbon is shown in figure A.1. The
parameters are listed in table A.1.
154
Table A.1: Parameters for the adsorption isotherm of ethane on active carbon.
Parameter Value Unit
Adsorption temperature
Tads
303.15 K
Critical temperature Tc of
ethane
305.4 K
Critical pressure pc of
ethane
48.8 bar
Critical pressure pc of
ethane
4880000 Pa
Hamaker constant Haj of
active carbon
6.0 E-20 J
Van-der-Waals diameter of
active carbon σj
3.4 E-10 m
Van-der-Waals diameter of
ethane σi
3.72 E-10 m
Minimum distance σi,j 3.56 E-10 m
Specific surface of the
adsorbent
1.1 E+6 m2/kg
Specific surface of the
adsorbent
1.1 E+3 m2/g
Toth isotherm parameter
nmax, equation (3.81)
14.46 mol/kg
Toth isotherm parameter
m, equation (3.81)
0.289 -
Toth isotherm parameter
1/KT, equation (3.81)
2.010 (kPa)m
The experimental values of the adsorption isotherm are compared with the adjusted
Toth isotherm and the predictive isotherm calculated with equation (A.8). The Toth
isotherm parameter 1/KT is given in the literature example in kPa [105], so the partial
pressure p in the Toth isotherm has to be inserted also in kPa.
155
Figure A.1: Adsorption isotherm of ethane on active carbon at Tads = 303.15 K.
Figure A.2: Adsorption isotherm of ethane on active carbon at Tads = 303.15 K in a
double logarithmic plot.
Ethane on active carbon
0
0,5
1
1,5
2
2,5
3
3,5
4
0 20000 40000 60000 80000 100000 120000
Adsorption pressure p in Pa
Load
ing
n in
mol
/kg
experimentToth isothermpredictive
Ethane on active carbon
1E-04
1E-03
1E-02
1E-01
1E+00
1E+01
1E+00 1E+01 1E+02 1E+03 1E+04 1E+05 1E+06Adsorption pressure p in Pa
Load
ing
n in
mol
/kg
experimentToth isothermpredictiveHenry predictiveSaturation predictive
156
With given loading n the partial pressure p of the adsorption isotherm can be
calculated by
( ) ( )He
kgmoln
barpKT
Sfnp c
cBET
=2exp
(A.9)
With equations (A.8) and (A.9) the adsorption isotherm of any adsorptive on activated
carbon can be calculated analytically in both directions.
Appendix B: Materials and Methods This study is based on both literature data and own adsorption experiments.
Experimental Henry coefficients were calculated by Langmuir and Toth isotherm
adaptations. Henry coefficients in the literature were taken from tables or calculated
by regressing the isotherm data with Langmuir and Toth parameters.
B.1. Materials
B.1.1. Experimental data
Adsorption experiments were conducted at the Lehrstuhl für Technische Chemie II of
the Technische Universität München in Garching and at the Institut für
Nichtklassische Chemie e.V. at the Universität Leipzig. In tables B.1 and B.2 the
used adsorptives and adsorbents and their relecant properties are listed.
157
Table B.1: Adsorptives used for adsorption isotherms with their properties.
Adsorptive Methanol (CH3OH) Dichloromethane (CH2Cl2)
Molar mass Mi in kg/mol [106] 0.032042 0.084933
Critical temperature Tc in K [106] 512.6 510
Critical pressure pc in Pa [106] 6300000 8090000
Normal boiling temperature Tb in K
[106]
337.7 313
Van-der-Waals diameter σj in m
(equation 5.19)
3.74*10-10 4.06*10-10
Critical molecule diameter in m 3.6*10-10 3.73*10-10
Dipole moment µ in debye (1 debye
= 3,34*10-30 Cm) [106]
1.7 1.8
Polarisability α in (C2*m2)/J [6] and
equation (3.67)
3.66*10-40 7.21*10-40
Quadrupole moment Q in Cm2 [107] 4.337*10-40 1.0*10-39
158
Table B.2: Adsorbents used for adsorption isotherms with their properties.
Adsorbent NaY MS5A NaX (13X)
Specific surface
SBET in m2/kg,
measured
655400 464135 521977
Micropore volume
vmicro in 10-3
m3/kg, measured
0.2723 0.2323 0.2643
Bulk density ρb in
kg/m3 [108, 24]
650 700 675
Bulk porosity
εb [24]
0.6 0.6 0.6
Apparent density
ρapp in kg/m3, ρapp
= ρb/εb
1100 1166 1125
Refractive index
n, equation (3.63)
1.61 1.58 1.63
Chemical
structure [24]
Na56
((AlO2)56(SiO2)106)
Ca5Na2
((AlO2)12(SiO2)12)
Na86
((AlO2)86(SiO2)106)
Effective pore
opening in m,
[108, 24]
10*10-10 5*10-10 10*10-10
Van-der-Waals
diameter σj in m,
table 5.2
4.2*10-10 3.8*10-10 3.9*10-10
Number density
of atoms ρj,at in
1/m3, equation
(3.92)
3.28*1028 3.31*1028 3.72*1028
micro
jBET
vS σ
1.01 0.76 0.75
159
The methanol had a purity of 99.9 mass-%. For experiments in Garching it was a
Merck product with the trade name Uvasol and is normally used for spectroscopy
purposes. Methanol adsorption experiments in Leipzig were conducted with the
Merck product LiChrosolv which is suitable for liquid chromatography purposes. The
dichloromethane had a purity of 99.8 mass-%. For experiments in Garching it was a
Sigma-Aldrich product with the trade name AMD Chromosolv and is normally used
for chromatography purposes. Dichloromethane adsorption experiments in Leipzig
were conducted with the Merck product pro analysi . The adsorbents MS5A and NaX
were from Grace Davison. MS5A has the trade name Sylobead MS C 522 and NaX
(13X) the trade name Sylobead MS C 544. The adsorbent NaY were from Akzo
Nobel.
Methanol is a high value chemical in many processes, e.g. the conversion to highly
knockproofed gasoline via zeolite catalysts. Dichloromethane is a solvent for resins,
fats and plastics. Both adsorptives have a medium dipole moment, high enough to
dominate with their dipole/charge potential the Hamaker potential and small enough
to get into the Henry region with a gravimetric experimental set-up when adsorbed on
zeolites. The zeolites MS5A, NaX and NaY are standard adsorbents in many
chemical processes like removal of pollutants.
B.1.2. Literature data
Literature data were collected and analyses was made for verifying the models
derived in this study [1, 3, 5, 7, 8, 9, 11, 12, 13, 18, 20, 22, 23, 25, 28, 29, 31, 33, 34,
35, 36, 37, 38, 39, 40, 45, 60, 91, 92, 115, 116, 117]. Henry coefficients were
tabulated in most of the cases, if not the adsorption isotherms were analysed by
using the Langmuir plot explained in Chapter 3 or with Langmuir parameter
regressing in the program Origin or with Toth isotherm regressing. The properties of
the adsorptives were found in Reid et al. [106] and in the CRC Handbook [6].
Especially quadrupole moments were found in the Computational Chemistry
Comparison and Benchmark Database [107] and in further literature [2, 14, 15, 16, 17,
42, 43, 103]. The examined adsorptives and adsorbents with their properties are
listed in appendix C and appendix D, respectively.
160
B.2. Experimental set-up
B.2.1. Adsorption isotherm measurements
The adsorption equilibrium isotherms were measured at the Lehrstuhl für Technische
Chemie 2 of the Technische Universität München gravimetrically in a modified
SETARAM TG-DSC 111 microbalance (see figure B.1) which is able to measure
differences in mass down to 10-6 g which is also the accuracy. The sample weight in
the microbalance was applied between 13 to 22 mg in order to record precise
measurement and better signal-to-noise ratio. The pressure inside the system can be
reduced down to 10-7 mbar with a Pfeiffer turbo molecular pump and the temperatures
can be applied within the range of 25°C to 750°C with an accuracy of 0.15°C. The high
vacuum was measured with a Pfeiffer TPG 300 pressure transducer. The experiments
were carried out in the pressure range from 10-3 to 13 mbar (0.1 Pa to 1300 Pa),
measured with a BARATRON 122 A pressure transducer. The transducer has an
accuracy of 0.5 % of reading [108]. For activation, the solid samples were heated in
vacuum (p < 10-6 mbar) to 500 °C with 10 °C/min and held 4 h at 500 °C. After
activation, the temperature was reduced to ca. 45°C. The liquid sorbates were added
in pulses to the high vacuum system in which they vaporised. After each pulse the
system was equilibrated. This was monitored by observation of the sample weight and
pressure. The system was regarded to be in steady state or equilibrium, when changes
in any of the parameters pressure, mass or heating flux were not observed any more
[110]. Figure B.2 shows the experimental set-up.
Figure B.1: SETARAM TG-DSC 111 microbalance.
161
Figure B.2: Experimental set-up for adsorption isotherm measurements at the Institut
für Technische Chemie 2 of the Technische Universität München.
The data for the adsorption isotherm graphs were calculated by the following method:
The adsorbents are hold by a crucible in the microbalance. Its mass after activating is
the so-called blank mass m0. The mass of the crucible with activated adsorbent is the
mass m1. After equilibrium of the injected adsorptive with the adsorbent the new
absolute mass mi+1 is measured. So one can calculate the equilibrium loading ni+1 of
every adsorption step with its equilibrium pressure could be measured by
11
0
111 >
−= +
+ iforMm
mmn
adsorptive
ii (B.1)
where the masses are in kg and Madsorptive is the molar mass of the adsorptive in
kg/mol. With equation (B.1) the equilibrium loading n in mol/kg regarding to the
equilibrium pressure p in Pa is obtained.
162
The adsorption equilibrium isotherms at the Institut für Nichtklassische Chemie e.V.
from the Universität Leipzig were measured gravimetrically with a high resolution
magnetic suspension balance from Rubotherm Präzisionsmesstechnik GmbH which
had a weighing instrument from Mettler-Toledo of type AT261. The sample weight in
the microbalance was applied between 10 to 30 g. The relative error of this
instrument is 0.002 % of the measured value and its resolution is 0.01 mg [113]. The
pressure inside the system can be reduced down to 10-3 mbar with a Pfeiffer turbo
molecular pump TMH 071P and the temperatures can be applied within the range of
25°C to 400°C with an accuracy of 0.15°C of the used PT100 sensor. The high
vacuum was measured with an MKS pressure transducer, Baratron Model 627B, and
the low vacuum with a Newport/Omega pressure transducer, Model SN2713. The
experiments were carried out in the pressure range from 10-1 mbar to 1 bar (10 Pa to
100.000 Pa). The transducers had an accuracy of 0.12 % of reading for the MKS
model and 0.03 % of the end value 1 bar for the Newport/Omega model [114]. For
activation, the solid samples were heated in vacuum (p < 10-3 mbar) to 370°C with 1
K/min and held 10 h at 370°C. After activation, the temperature was reduced to 323
K. The liquid sorbates were degassed after condensing with liquid nitrogen with a
vacuum pump from ATS Antriebstechnik GmbH. This was done several times. After
this the sample was added in pulses to the high vacuum system in which they
vaporised. After each pulse the system was equilibrated. This was monitored by
observation of the sample weight and pressure. The system was regarded to be in
steady state or equilibrium, when changes in any of the parameters were not
observed. Figure B.3 shows the experimental set-up.
Figure B.3: Experimental set-up for adsorption isotherm measurements in Leipzig.
163
The data for the adsorption isotherm graphs were calculated by the same method as
described above for equation (B.1).
In general the first set-up was taken for high-vacuum measurements from 0.1 Pa to
100 Pa and the second set-up for medium-vacuum measurements from 100 Pa to
saturation pressure of the adsorptive at room temperature.
B.2.2. BET surface and micropore volume measurements
BET surface measurements were conducted with nitrogen at 77 K as described in
Chapter 3. The micropore volume of the adsorbents were derived from t-plot
analysis. Results are listed in table B.2.
164
B.2
.3. A
dsor
ptiv
e da
ta
Tabl
e B
.4: E
xam
ined
ads
orpt
ives
and
thei
r rel
evan
t pro
perti
es.
Ads
orpt
ive
Mol
ar m
ass
in
10-3
kg/
mol
Va
n-de
r- W
aals
di
amet
er σ
i in m
Dip
ole
mom
ent m
iin
deby
e P
olar
isab
ility
αi
in (C
2 *m2 )/J
Q
uadr
upol
e m
omen
t Qi in
Cm
2C
ritic
al
tem
pera
ture
Tc i
n K
Crit
ical
pre
ssur
e p c
in
Pa
N2
28.0
13.
13E
-10
01.
93E
-40
5.07
E-4
012
6.2
3390
000
CO
28
.01
3.15
E-1
00.
12.
14E
-40
9.87
E-4
013
2.9
3500
000
CH
4 16
.04
3.24
E-1
00
3.00
E-4
00
190.
446
0000
0O
2 31
.99
2.93
E-1
00
1.76
E-4
02.
70E
-40
154.
650
4000
0K
r 83
.80
3.15
E-1
00
2.76
E-4
00
209.
455
0000
0C
2H4
28.0
53.
59E
-10
04.
73E
-40
1.44
E-3
928
2.4
5040
000
CH
F 3
70.0
13.
70E
-10
1.6
3.95
E-4
01.
07E
-39
299.
348
6000
0C
3H6
42.0
84.
15E
-10
0.4
6.91
E-4
01.
80E
-40
364.
946
0000
0N
H3
17.0
33.
09E
-10
1.5
3.13
E-4
08.
67E
-40
405.
511
3500
00H
2 2.
016
2.76
E-1
00
8.93
E-4
11.
33E
-40
33.2
1300
000
CO
2 44
.01
3.24
E-1
00.
72.
89E
-40
1.50
E-3
930
4.2
7380
000
C2H
6 30
.07
4.78
E-1
00
4.98
E-4
02.
47E
-40
305.
448
8000
0C
3H8
44.0
94.
15E
-10
07.
03E
-40
1.80
E-4
036
9.8
4250
000
C4H
10 (n
-But
ane)
58
.12
4.52
E-1
00
9.06
E-4
02.
67E
-40
425.
238
0000
0C
4H10
(i-B
utan
e)
58.1
24.
52E
-10
0.1
9.06
E-4
02.
67E
-40
408.
236
5000
0C
6H14
n-H
exan
e 88
.18
5.18
E-1
00
1.32
E-3
90
507.
530
1000
0C
6H12
(cyc
lohe
xane
) 84
.16
4.82
E-1
00
1.22
E-3
93.
00E
-40
553.
540
7000
0C
6H14
(2m
ethy
lpen
tane
) 88
.18
5.14
E-1
00.
11.
31E
-39
2.67
E-4
049
7.5
3010
000
SF 6
14
6.05
4.12
E-1
00
7.28
E-4
00
319
3760
000
CH
2Cl 2
84.9
34.
06E
-10
1.8
7.21
E-4
01.
00E
-39
510
6030
000
C2H
Cl 3
131.
394.
54E
-10
0.9
1.12
E-3
96.
67E
-40
572
5050
000
He
4.0
2.66
E-1
00
2.20
E-4
10
5.19
2270
00
165
Tabl
e B
.4: E
xam
ined
ads
orpt
ives
and
thei
r rel
evan
t pro
perti
es.
Ads
orpt
ive
Mol
ar m
ass
in
10-3
kg/
mol
Va
n-de
r- W
aals
di
amet
er σ
i in m
Dip
ole
mom
ent m
iin
Deb
ye
Pol
aris
abili
ty α
i in
(C2 *m
2 )/J
Qua
drup
ole
mom
ent Q
i in C
m2
Crit
ical
te
mpe
ratu
re T
c in
KC
ritic
al p
ress
ure
p c in
P
a C
6H6 (
benz
ene)
78
.11
4.56
E-1
00
1.11
E-3
91.
20E
-39
562.
248
9000
0C
2H4 (e
then
e)
28.0
53.
59E
-10
04.
73E
-40
1.44
E-3
928
2.4
5040
000
CH
3OH
(met
hano
l) 32
.04
3.74
E-1
01.
73.
66E
-40
4.34
E-4
051
2.6
8090
000
C8H
10 (p
-xyl
ene)
10
6.16
5.25
E-1
00
1.57
E-3
92.
56E
-39
616.
235
1000
0C
10H
8 (na
phta
lene
) 12
8.17
5.34
E-1
00
1.94
E-3
94.
50E
-39
748.
440
5000
0A
r 39
.95
2.94
E-1
00
1.83
E-4
00
150.
848
7000
0H
FC13
4a (C
2H2F
4)
102.
034.
23E
-10
2.1
4.87
E-4
04.
13E
-39
374
4060
000
B.2
.4. A
dsor
bent
dat
a
Tabl
e B
.5: E
xam
ined
ads
orbe
nts
from
the
liter
atur
e an
d th
eir r
elev
ant p
rope
rties
. A
dsor
bent
C
hem
ical
stru
ctur
e σ j
in
10-1
0 m
Num
ber
n a
of a
tom
s in
the
solid
mol
ecul
e
Den
sity
ρ app
of
the
solid
in
kg/m
3
Mol
ecul
ar
mas
s M
j of
the
solid
in
kg/m
ol
Num
ber
dens
ity
ρ j,a
t of
atom
s in
1/m
3
Ref
ract
ive
inde
x of
the
solid
Aver
age
pola
risab
ility
of
the
solid
atom
s j
α in
C2 m
2 /J
Aver
age
num
ber
of
vale
nce
elec
tron
bind
ings
s
Ham
aker
cons
tant
Ha j
in J
MS
4A
Na 1
2((A
lO2)
12(S
iO2)
12)
3.80
84
16
80
1.70
5.
00E
28
1.44
1.
40E
-40
2.30
6.
50E
-20
MS
5A
Ca 5
Na 2
((AlO
2)12
(SiO
2)12
) 3.
80
79
1150
1.
68
3.27
E28
1.
57
2.67
E-4
0 2.
43
7.66
E-2
0
NaX
N
a 86(
(AlO
2)86
(Si0
2)10
6)
3.86
66
2 11
00
12.0
3.
64E
28
1.62
2.
55E
-40
2.32
8.
70E
-20
NaY
N
a 56(
(AlO
2)56
(SiO
2)13
6)
4.
20
632
1100
12
.76
3.28
E28
1.
613
2.82
E-4
0 2.
43
8.40
E-2
0
Sili
calit
e-1
Si96
O19
2 3.
85
298
1760
5.
80
5.58
E28
1.
46
1.32
E-4
0 2.
67
7.89
E-2
0
166
Appendix C: Alternative way for deriving the potential factor fquadcha In Section 5.1 the assumption has been made that no permanent dipole or
quadrupole in an adsorbent molecule is existing and inSection 5.2.1 it has been
stated that the time avareged dipole moment is zero. However, a strong quadrupole
moment Qi with a certain structure can induce a dipole moment µi in an adsorbent
molecule. The ratio between the moments µi and Qi of an adsorptive molecule can be
derived when equation (5.57) is divided by equation (5.51):
( ) 202,
2
2
=+ z
Q
jii
i
σµ (C.1)
or for z = 0 (maximum value of Qi for a given µi)
jiiiQ ,20 σµ= (C.2)
The only adsorptive molecule with a strong quadrupole moment but a rather dipole
moment already often investigated with respect to adsorption isotherms is CO2 (Qi =
15*10-40 Cm2, µi = 2.34*10-30 Cm). In the previous sections the potential factors fdipcha
= findcha = 1.35 jSi
Al
have been found. These factors are supported by experimental
data. Let us assume that the potential factor fquadcha is also 1.35*jSi
Al
. An
agreement between experimental results for the system CO2-MS4A is given if the
dimensionless energy ads
quadcha
kTΦ
− in equation (5.57) is multiplied by the factor 8 which
leads to fquadcha = 1,35/8 *jSi
Al
=0.168*
jSiAl
. As a first approach it is further
assumed that the induced dipole moment µj (permanent due to the close
neighborship of an quadrupole moment Qi) is related. Now an effective quadrupole
moment Qi,eff is introduced which takes into account the interaction between Qi and µi
With σi,j = 4.05*10-10 m and µi = µj we find
167
240,, 10*4.4220 CmQ jiieffi
−== σµ (C.3)
and finally
810*0.1510*4.42
2
40
402, =
=
−
−
i
effi
(C.4)
It is stressed here that the hypothesis here presented should be tested by
experimental results, for instance by experimental data of the system C2H2F4/zeolite.
168
References
[1] Azarenkov, A.N., Al�tshuler, G.B., Belashenkov, N.R. and S.A. Kozlov.
1993. Fast nonlinearity of the refractive index of solid-state dielectric active
media. Quantum Electronics 23:633-655
[2] Benavides, A.L., Guevara, Y. and A.F. Estrada-Alexanders. 2000. A
theoretical equation of state for real quadrupolar fluids. Journal of
Chemical Thermodynamics 32:945�951.
[3] Bieske, E. J., Nizkorodov, S.A., Bennett, F.R. and J.P. Maier. 1995.
Combined infrared and ab initio study of the H2-HN2+ complex.
International Journal of Mass Spectrometry and Ion Processes 149/150:.
167�177.
[4] Braun, Björn. 2004. Aggregation and Agglomeration in Percipitative
Crystalization. Dissertation at the Technical University of Munich.
[5] Choudhary, V. R. and S. Mayadevi. 1996. Adsorption of methane, ethane,
ethylene and carbon dioxide on silicalite-1. Zeolites 17:501-507.
[6] Lide, David R., Editor-in-Chief. 2004-2005. CRC Handbook of Chemistry
and Physics, 85th Edition. London, New York, Washington: CRC Press.
[7] Dunne, J. A., Rao, M., Sircar, S., Gorte, R.J. and A.L. Myers. 1996.
Calorimetric Heats of Adsorption and Adsorption Isotherms. 2. O2, N2, Ar,
CO2, CH4, C2H6 and SF6 on NaX, H-ZSM-5, and Na-ZSM-5 Zeolites.
Langmuir 12:5896-5814.
[8] Markmann, Birgitt. 1999. Zur Vorhersagbarkeit von
Adsorptionsgleichgewichten mehrkomponentiger Gemische. Dissertation
at the Technical University of Munich.
169
[9] Fox, J.P., Rooy, V. and Simon P. Bates. 2004. Simulating the adsorption
of linear, branched and cyclic alkanes in silicalite-1 and AIPO4-5.
Microporous and Mesoporous Materials 69:9-18.
[10] Fried, J. R. and N. Hu. 2003. The molecular basis of CO2 interaction with
polymers containing fluorinated groups. Polymer 44:4363�4372.
[11] Giaya, A., and Robert W. Thompson. 2002. Single-component gas phase
adsorption and desorption studies using a tapered element oscillating
microbalance. Microporous and Mesoporous Materials 55:265-274.
[12] Grande, C.A., Gigola, C. and A.E. Rodrigues. 2002. Adsorption of Propane
and Propylene in Pellets and Crystals of 5A Zeolite. Industrial and
Engineering Chemistry Research 41:85-92.
[13] Grande, C.A., Gigola, C. and A.E. Rodrigues. Propane-Propylene Binary
Adsorption on Zeolite 4A. Adsorption 9(4):321-329.
[14] Greenhow, C. and W.V. Smith. 1951. Molecular Quadrupole Moments of
N2 and O2. The Journal of Chemical Physics 10:1298-1230.
[15] Halkier, A. and S. Coriani. 1999. On the molecular electric quadrupole
moment of C2H2. Chemical Physics Letters 303:408�412.
[16] Heard, G. L. and R.J. Boyd. 1997. Density functional theory studies of the
quadrupole moments of benzene and naphthalene. Chemical Physics
Letters 277:252�256.
[17] Heard, G. L. and R.J. Boyd. 1997. Calculation of Quadrupole Moments of
Polycyclic Aromatic Hydrocarbons: Applications to Chromatography,
Journal of Physical Chemistry A 101:5374-5377.
[18] Helminen, J., Helenius, J. and E. Paatero. 2001. Adsorption Equilibria of
Ammonia Gas on Inorganic and Organic Sorbents at 298,15 K. Journal of
chemical and engineering data, 46:391-399.
170
[19] Hirschfelder, Joseph O., Curtiss, Charles F. and Bird R. Byron, eds. 1964.
Molecular Theory of Gases and Liquids., New York: John Wiley & Sons,
Inc.
[20] Inel, O., Topaloglu, D., Askin, A. and F. Tümsek. 2002. Evaluation of the
thermodynamic parameters for the adsorption of some hydrocarbons on
4A and 13X zeolites by inverse gas chromatography. Chemical
Engineering Journal 88:255-262.
[21] Israelachvili, Jacob N. 1991. Intermolecular and Surface Forces, 2nd
Edition. San Diego, San Francisco, New York: Academic Press.
[22] Jaramillo, E. and M. Chandross. 2004. Adsorption of Small Molecules in
LTA Zeolites. 1. NH3, CO2, and H2O in Zeolite 4A. The Journal of Physical
Chemistry. B, Condensed Matter, Materials, Surface, Interfaces &
Biophysical Chemistry. 108(52):20155-20159.
[23] Vashchenko, L.A. and V.V. Katal´nikova. 1996. Adsorption of helium on
zeolite NaA. Russian Chemical Bulletin 45(5):1230-1231.
[24] Kast, Werner. 1998. Adsorption aus der Gasphase. Weinheim, Germany:
VCH Press.
[25] Klemm, E., Wang, J. and G. Emig. A comparative study of the sorption of
benzene and phenol in silicalite, HAlZSM-5 and NaAlZSM-5 by computer
simulation. Microporous and Mesoporous Materials 26:11-21.
[26] Maurer, Stephan. 2000. Prediction of Single-Component Adsorption
Equilibria. Dissertation at the Technical University of Munich.
[27] Mersmann, A., Fill, B. and S. Maurer. 2002. Single and Multicomponent
Adsorption Equilibria of Gases. Fortschritt-Berichte VDI, Reihe 3
Verfahrenstechnik 735.
[28] Moeller, A. and R. Staudt. 2005. Personal communication.
171
[29] Mohr, R. J., Vorkapic, D., Rao, M.B. and S. Sircar. 1999. Pure and Binary
Gas Adsorption Equilibria and Kinetics of Methane and Nitrogen on a 4A
Zeolite by Isotope Exchange Technique. Adsorption 5:145-158.
[30] Munataka, K., Fukumatsu, T., Yamatsuki, S., Tanaka, K. and M.
Nishikawa. 1999. Adsorption Equilibria of Krypton, Xenon, Nitrogen and
Their Mixtures on Molecular Sieve 5A and Activated Charcoal. Journal of
Nuclear Science and Technology 36:818-829.
[31] Nam, G., Jeong, B., Kang, S., Lee, B. and D. Choi. 2005. Equilibrium
Isotherms of CH4, C2H6, C2H4, N2 and H2 on Zeolite 5A Using a Static
Volumetric Method. Journal of Chemical & Engineering Data 50:72-76.
[32] Palik, Edward D. 1991. Handbook of Optical Constants of Solids I, II, III.,
New York: Academic Press, Inc.
[33] Rep, M., Palomares, A.E., Eder-Mirth, G., van Ommen, J.G., Rösch, N.
and J.A. Lercher. 2000. Interaction of Methanol with Alkali Metal
Exchanged Molecular Sieves. 1. IR Spectroscopic Study. Journal of
Physical Chemistry B 104:8624-8630.
[34] Ruthven, D.M. and B.K. Kaul. 1993. Adsorption of Aromatic Hydrocarbons
in NaX Zeolite. 1. Equilibrium. Industrial & Engineering Chemistry
Research 32:2047-2052.
[35] Sander, Stefan. 2000. Experimentelle Bestimmung von
Adsorptionsgleichgewichten und Evaluation von
Adsorptionsgleichgewichtsmodellen. Dissertation at the Carl von.
Ossietzky University, Oldenburg.
[36] Schweighart, Peter. 1994. Adsorption mehrerer Komponenten in
biporösen Adsorbentien. Dissertation at the Technical University of
Munich.
[37] Sievers, Werner. 1993. Über das Gleichgewicht der Adsorption in Anlagen
zur Wasserstoffgewinnung. Dissertation at the Technical University of
Munich.
172
[38] Sieperstein, F.R., Gorte, R.J. and A.L. Myers. 1999. Measurement of
Excess Functions of Binary Gas Mixtures Adsorbed in Zeolites by
Adsorption Calorimetry. Adsorption 5:8-12.
[39] Sieperstein, F.R. and A.L. Myers. 2001. Mixed-Gas Adsorption. AIChE
Journal 47(5):1141-1159.
[40] Talu, O. and A.L. Myers. 2001. Reference potentials for adsorption of
helium, argon, methane, and krypton in high-silica zeolites. Colloids and
Surfaces A: Physicochemical and Engineering Aspects. 187:83-93.
[41] Van der Hoeven, Ph.C. 1991. Electrostatic Stabilization of Suspensions in
Non-Aqueous Media. Dissertation at the Landbouwuniversiteit te
Wageningen.
[42] Vrabek, J., Stoll, J. and H. Hasse. 2001. A Set of Molecular Models for
Symmetric Quadrupolar Fluids. Journal of Physical Chemistry 105:12126-
12133.
[43] Wan, Q., Ramaley, L. and R. Guy. 1997. Quadrupolar Effects on the
Retention of Aromatic Hydrocarbons in Reversed-Phase Liquid
Chromatography. Analytical Chemistry 69:4581�4585.
[44] Xu, K., Li, Y. and Liu, W. 1998. Application of perturbation theory to chain
and polar fluids Pure alkanes, alkanols and water. Fluid Phase Equilibria,
142:55-66.
[45] Zhang, Y., Furukawa, S. and T. Nitta. 2004. Monte Carlo Simulation
Studies on Adsorption of Propane/Propylene in NaX Zeolite. Journal of
Chemical Engineering of Japan 37(4):563-568.
[46] Zhu, W., Kapteijn, F. and J.A. Moulijn. 2000. Equilibrium Adsorption of
Light Alkanes in Silicalite-1 by the Inertial Microbalance Technique.
Adsorption 6:159-167.
[47] Mersmann, A., Kind, M. and Johann Stichlmair. 2005. Thermische
Verfahrenstechnik. Berlin Heidelberg: Springer Verlag.
173
[48] Ruthven, Douglas M. 1984. Principles of Adsorption and Adsorption
Processes. New York: John Wiley & Sons, Inc.
[49] Götzinger, Martin. 2004. Zur Charakterisierung von Wechselwirkungen
partikulärer Feststoffoberflächen. Dissertation at the Technical University
of Munich.
[50] Mehler, Carsten. 2003. Charakterisierung heterogener
Feststoffoberflächen durch Erweiterung eines dielektrischen
Kontinuumsmodells. Dissertation at the Technical University of Munich.
[51] Groth, Thomas. 2002. Die Bedeutung der Volumen- und
Oberflächeneigenschaften von Biomaterialien für die Adsorption von
Proteinen und nachfolgende zelluläre Reaktionen. State doctorate at the
University of Potsdam.
[52] Dabrowski, A. 2001. Adsorption - from theory to practice. Advances in
Colloid and Interface Science 93:135-224.
[53] Matisova, E. and S. Skrabakova. 1995. Carbon sorbents and their
utilization for the preconcentration of organic pollutants in environmental
samples. Journal of Chromatography A 707:145-179.
[54] Yang, T and A. C. Lua. 2003. Characteristics of activated carbons
prepared from pistachio-nut shells by physical activation. Journal of Colloid
and Interface Science 267:408-417.
[55] Weitkamp, J. 2000. Zeolites and catalysis. Solid State Ionics 131:175-188
[56] Röhr, Caroline. 2005. Siedende Steine. Recitation Wissenschaftliche
Gesellschaft. Internet site: http://ruby.chemie.uni-freiburg.de/Vorlesung/
PDF_Demos/zeolithe.pdf.
[57] Staudt, Reiner. 1994. Analytische und experimentelle Untersuchungen von
Adsorptipnsgleichgewichten von reinen Gasen und Gasgemischen an
Aktivkohlen und Zeolithen. Dissertation at the Universität-
Gesamthochschule-Siegen.
174
[58] Simon, Ulrich. 2000. Innere Werte Zeolithe: nanoporöse Materialien für
vielfältige Anwendungsfelder. Essener Unikate 13:66-77.
[59] Kalipcilar, H. and A. Culfaz. 2000. Synthesis of Submicron Silicalite-1
Crystals from Clear Solutions. Cryst. Res. Technol. 35(8):933-942.
[60] Zhu, W., Kapteijn, F. and J.A. Moulijn. 2000. Adsorption of light alkanes on
silicalite-1 : Reconciliation of experimental data and molecular simulations.
Physical Chemistry Chemical Physics 2:1989-1995.
[61] Jentys, A. and J.A. Lercher. 2001. Chapter8: Techniques of zeolite
chracterization. Studies in Surface Science and Catalysis 137:345-386.
[62] Brunauer, S., Emmett, P.H. and E. Teller. 1937. Adsorption of Gases in
Multimolecular Layers. Journal of the American Chemical Society 60:309-
316.
[63] Lercher, Johann A. 1999. Chapter 13: Adsorption methods for the
assessment of the specific surface area, the pore size distribution and the
active sites of heterogeneous catalysts. Catalysis: An Integrated
Approach. Editors Van Santen, R.A. et al.
[64] Fletcher, Ashleigh. 2006. Porosity and sorption behaviour. Internet site:
http://www.staff.ncl.ac.uk/a.j.fletcher.
[65] Westermarck, Sari. 2000. Use of mercury porosimetry and nitrogen
adsorption in characterisation of the pore structure of mannitol and
microcrystalline cellulose powders, granules and tablets. Dissertation at
the University of Helsinki.
[66] Washburn, E.W. 1921. The dynamics of capillary flow. Physical Review
17(3):273-283.
[67] Micromeritics. 2005. Data reduction methods. Internet site
http://www.micromeritics.com
175
[68] Lowell, S., Shields, J.E., Thomas, M.A. and M. Thommes. 2004.
Characterization of Porous Solids and Powders: Surface Area, Pore Size
and Density. Particle Technology Series Volume 16.
[69] Quantachrome. 2003. Dichtebestimmung von Feststoffen und neue
Anwendungen mit dem ULTRAPYCNOMETER 1000T. Partikelwelt,
Aktuelle Fachbeiträge der Quantachrome GmbH 2:7-8.
[70] Viana, M., Jouannin, P., Pontier, C. and D. Chulia. 2002. About
pycnometric density measurements. Talanta 57:583-593.
[71] Zundu, L. and H. Yidong. 2004. Qualitative relationship between refractive
index and atomic parameters of solid materials. Journal of Rare Earths
22(4):486-488.
[72] Marler, B. 1988. On the relationship between refractive index and density
of SiO2-polymorphs. Physics and chemistry of minerals 16:286-290.
[73] Susa, M., Kamijo, Y., Kusano, K. and R. Kojima. 2005. A predictive
equation for the refractive indices of silicate melts containing alkali, alkaline
earth and aluminium oxides. Glass Technology 46(2):55-61.
[74] Striebel, F. 2005. Physikalisch-Chemisches Praktikum für
Fortgeschrittene. Versuch F32: Dielektrizitätskonstante, Dipolmoment und
Molrefraktion. Institut für Physikalische Chemie Universität Karlsruhe.
[75] Hamaker, H.C. 1937. The London-Van der Waals attraction between
spherical particles. Physica 4(10):1058-1072.
[76] Ackler, H.D., French, R.H. and Y. Chiang. 1995. Comparisons of Hamaker
Constants for Ceramic Systems with Intervening Vacuum or Water: From
Force Laws and Physical Properties. Journal of Colloid and Interface
Science 179:460-469.
[77] Bergström, L. Hamaker constants of inorganic materials. 1997. Advances
in Colloid and Interface Science 70:125-169.
176
[78] Brunauer, S., Deming, L.S., Deming, W.E. and E. Teller. 1940. Journal of
the American Chemical Society 62:1723-.
[79] Czepirski, L., Balys, M.R. and E. Komorowska-Czepirska. 2000. Some
generalization of Langmuir adsorption isotherm. Internet Journal of
Chemistry 3(14).
[80] Langmuir, I. 1918. The adsorption of gases on plane surfaces of glass,
mica and platinum. Journal of the American Chemical Society 40:1361-
1402.
[81] Yang, Ralph T. 1987. Gas separation by adsorption processes. Stoneham:
Butterworth Publishers.
[82] Kajszika, Holger. 1998. Adsorptive Abluftreinigung und
Lösungsmittelrückgewinnung durch Inertgasregenerierung. Dissertation at
the Technical University of Munich.
[83] Toth, J. 1971. State equations of the solid-gas interface layers. Acta Chim.
Acad. Sci. Hung. 69:311-328.
[84] Polanyi, M. 1916. Adsorption von Gasen (Dämpfen) durch ein festes
nichtflüchtiges Adsorbens. Verhandlungen der Deutschen Physikalischen
Gesellschaft 18(4):55-80.
[85] Dubinin, M.M. 1967. Adsorption in Micropores. Journal of Colloid and
Interface Science 23:487-499.
[86] Dubinin, M.M. 1960. The Potential Theory of Adsorption of Gases and
Vapors for Adsorbents with energetically nonuniform surfaces. Chem. Rev.
60:235-241.
[87] Jentz, M.A. 2005. Molecular Modeling of Separations in Silicalite and
Crystalline Silicotitanate Systems. Research Thesis at the University of
Notre Dame, Department of Chemical and Biomolecular Engineering in
Hall.
177
[88] Smit, B. and R. Krishna. 2001. Monte Carlo simulations in zeolites. Current
Opinion in Solid State and Materials Science 5:455-461.
[89] Brandani, F., Ruthven, D. and C.G. Coe. 2003. Measurement of
Adsorption Equilibria by the Zero Length Column (ZLC) Technique Part 1:
Single-Component Systems. Industrial & engineering chemistry research
42(7):1451-1461.
[90] Brandani, F., Ruthven, D. and C.G. Coe. 2003. Measurement of
Adsorption Equilibria by the Zero Length Column (ZLC) Technique Part 2:
Binary Systems. Industrial & engineering chemistry research 42(7):1462-
1469.
[91] Harlick, P.J.E. and F.H. Tezel. 2003. Use of Concentration Pulse
Chromatography for Determining Binary Isotherms: Comparison with
Statically Determined Binary Isotherms. Adsorption 9:275-286.
[92] Harlick, P.J.E. and F.H. Tezel. 2004. An experimental adsorbent screening
study for CO2 removal from N2. Microporous and Mesoporous Materials
76:71-79.
[93] Levenspiel, Octave. 1972. Chemical Reaction Engineering Second
Edition. New York: John Wiley & Sons, Inc.
[94] Dilger, H. 2006. Physikalisch-Chemisches F-Praktikum, Dipolmoment.
Praktikum at the University of Stuttgart, Institut für Physikalische Chemie.
[95] Kuchling, H. 1986. Taschenbuch der Physik. Thun and Frankfurt/Main:
Verlag Harri Deutsch.
[96] Knerr, R. 1999. Goldmann Lexikon, Physik, Vom Atom zum Universum.
München: Wilhelm Goldmann Verlag.
[97] Borisover, M.D. and E.R. Graber. 1997. Specific interactions of organic
compounds with soil organic carbon. Chemosphere 34(8):1761-1776.
178
[98] van Oss, C.J., Chaudry, M.K. and R.J. Good. 1988. Interfacial Lifshitz-van
der Waals and polar interactions in macroscopic systems. Chem. Rev.
88:927-941.
[99] Salem, L. 1960. The calculation of dispersion forces. Molecular Physics
3:441-452.
[100] Slater, J.C. and J.G. Kirkwood. 1931. The Van-der-Waals forces in gases.
The physical review 37(6):682-697.
[101] Bloeß, Harald. 2001. Erzeugung und Charakterisierung von SiO2
Nanostrukturen auf Silizium. Dissertation at the Heinrich-Heine-University
Düsseldorf.
[102] Pierotti, R.A. and H.E. Thomas. 1971. Physical adsorption: The interaction
of gases with solids. Surface and Colloid Science 4:93-259.
[103] Halkier, A. and S. Coriani. 2001. State-of-the-art ab initio calculations of
the molecular electric quadrupole moments of hydrogen fluoride. Chemical
Physics Letter 346:329-333.
[104] Rechid, Juan. 2000. Elektrische Mikrocharakterisierung von
elektrochemisch hergestellten CIS-Solarzellen mittels EBIC. Dissertation
at the University of Oldenburg.
[105] Valenzuela, D.P. and A.L. Myers. 1989. Adsorption Equilibrium Data
Handbook. New Jersey: Prentice Hall, Englewood Cliffs.
[106] Reid, R.C., J.M. Prausnitz and B.E. Polling. 1988. The Properties of Gases
& Liquids. Singapore: McGraw-Hill Book Co.
[107] National Institute of Standards and Technology. 2006. Computational
Chemistry Comparison and Benchmark Database. http://www.nist.gov/
and http://srdata.nist.gov/cccbdb/.
[108] Grace Davison. 2006. Product Information Sheets for MS5A (Sylobead MS
C 522) and 13X (Sylobead MS C 544).
179
[109] University of Wisconsin. 2006. Range of alphas. Advances Laboratory,
Physics 407.
[110] Sievers, C. 2006. Personal communication: Experimental set-up for
adsorption isotherm measurements at the Department for Technical
Chemistry 2 at the Technical University of Munich.
[111] Al-Baghli, N.A. and K.F. Loughlin. 2006. Pure and Mixture Gas Adsorption
of Light Hydrocarbon on ETS-10 Zeolite. Chemical Engineering
Department, King Fahd University of Petroleum & Minerals.
[112] Klank, D. 2005. Oberflächen- und Porencharakterisierung mit Hilfe von
Gassorption und Quecksilberporosimetrie, Teil 5: Die Verwendung
alternativer Messgase bei Gassorptionsmessungen. Partikelwelt, Aktuelle
Fachbeiträge der Quantachrome GmbH 5:8-11.
[113] Lösch, W. Magnetic Suspension Balances. Rubotherm
Präzisionsmesstechnik GmbH.
[114] Möller, A. 2006. Personal communication. Institut für Nichtklassische
Chemie, Universität Leipzig.
[115] Talu, O., Zhang, S. and D.T. Hayhurst. 1993. Effect of Cations on Methane
Adsorption by Nay, MgY, Cay, SrY, and BaY Zeolites. Journal of Physical
Chemistry 97:12894-12898.
[116] Bellat, J., Simonot-Grange, M. and S. Jullian. 1995. Adsorption of gaseous
p-xylene and m-xylene on NaY, KY, and BaY zeolites: Part 1, Adsorption
equilibria of pure xylenes. Zeolites 15:124-230.
[117] Choudhary, V.R., Mayadevi, S. and A.P. Singh. 1995. Sorption isotherm of
methane, ethane, ethene and carbon dioxide on NaX, NaY and Na-
mordenite zeolites. Journal of the Chemical Society, Faraday Transactions
I 91(17):2935-2944.
180
[118] Morishige, K., Kittaka, S. and S. Ihara. 1985. Adsorption of methylamines
on dehydrated NaX, NaY and KY zeolites. Journal of the Chemical
Society, Faraday Transactions I 81:2525-2539.
[119] Nitta, T., Suzuki, T. and T. Katayama. 1991. Gas-Phase Adsorption
Equilibria for Acetone, Diethylether, methanol, and water on activated
carbon fiber. Journal of Chemical Engineering of Japan 24(2):160-165.
[120] Klank, D. 2005. Oberflächen- und Porencharakterisierung mit Hilfe von
Gassorption und Quecksilberporosimetrie, Teil 5: Die Verwendung
alternativer Messgase bei Gassorption. Partikelwelt, Aktuelle Fachbeiträge
der Quantachrome GmbH (5):8-13.
[121] Suk, Y. 2000. Characterization of the Physical, Thermal and Adsorption
Properties of a Series of Activated Carbon Fiber Cloths. M. S. Thesis,
University of Illinois at Urbana�Champaign.
[122] Mersmann, A. 2006. Personal Communication. (TU München, Germany).
[123] Laurin, M. 2005. Diffusion und Fluktuationen bei der CO-Oxidation an Pd-
getraegerten Nanopartikeln. FU Berlin Digitale Dissertation,
http://www.diss.fu-berlin.de/2006/13/index.html.
[124] B.P. Russell and M.D. LeVan. 1997. Coadsorption of Organic Compounds
and Water Vapour on BPL Activated Carbon. 3. Ethane, Propane, and
Mixing Rules. Ind. Eng. Chem. Res. 36:2380-2389.
[125] R. W. Morrison. Overview of Current Collective Protection Filtration
Technology. Guild Associates Inc.Baltimore, Maryland.
[126] F. Eder and J.A. Lercher. 1997. Alkane sorption in molecular sieves: the
contribution of ordering intermolecular interactions, and sorption on
Bronsted acid sites. Zeolites 18:75-81.
[127] Carbon Materials Research Group from the Center of Applied Energy
Research from the University of Kentucky. 2006. Historical production and
181
use of carbonic materials. Internet site http://www.caer.uky.edu/
carbon/history/carbonhistory.shtml.
[128] V.L. Lorenz. 1880. Annualen der Physik 11:70.
[129] H.A. Lorentz. 1880. Annualen der Physik 9:641.
[130] M.M. Dubinin and V.V. Astakhov. 1971. Description of Adsorption
Equilibria of Vapors on Zeolites over Wide Ranges of Temperature and
Pressure. Advances in Chemistry Series 102:69-85.
[131] U. Münstermann. 1984. Adsorption von CO2 aus Reingas und aus Luft am
Molekularsieb-Einzelkorn und im Festbett - Berücksichtigung von
Wärmeeffekten und H2O-Präadsorption. Dissertation at the Technical
University of Munich.
[132] Gerthsen, Kneser, Vogel. 1977. Gerthsen Physik. Berlin Heidelberg New
York Tokio: Springer.
[133] D. Pan. 1992. Untersuchungen zur Adsorption von Gasen am Zeolith
ZSM-5. Dissertation at the Technical University of Munich.